Benchmarks · /benchmarks/vals-ai-gpqa

GPQA Diamond

Name: GPQA Diamond
Creator: Vals AI

GPQA Diamond result as reported by Vals AI.

Source · Vals AI
Version · vals-ai snapshot 2026-10-03
Scores · 125

Test details

Visible tradeoffsThis is an objective signal, so it is mainly about measurable task performance rather than public taste.

source

Vals AI

metric

Accuracy (%)

judge

Objective

direction

higher better

group id

vals_gpqa_current

domain

Reasoning / math / science

What it measures vs what it misses

✓ Measures

Graduate-level Google-proof scientific reasoning.

✗ Misses

Adjacent skills outside the benchmark task mix, latency, and cost.

Why this countsIt is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.Same-test ruleThis percentile only compares models inside the exact benchmark/version group shown here. It is not a universal score.What it missesIt still misses product usability, latency, and whether the model stays correct in messy real workflows.

Leaderboard · this benchmark version

#1 · Gemini 3.1 Pro Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.1-pro-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 100%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

95.5%93.4% - 97.5%

#2 · GPT-5.6 Sol

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.6-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 99.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

95.2%93.1% - 97.3%

#3 · grok-4.6-high

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 98.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

94.7%92.5% - 96.9%

#4 · gemini-3.8-flash-high

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.8-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 97.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

94.4%91.5% - 97.3%

#5 · gemini-3.7-flash-high

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.7-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 96.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

93.9%91% - 96.9%

#6 · qwen3.8-max

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.8-max

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 96%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

93.7%91.3% - 96.1%

#7 · gemini-3.6-flash-high

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.6-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 95.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

93.4%90.8% - 96%

#8 · Claude Opus 5 (Adaptive Reasoning, Max Effort)

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 95.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

93.4%91% - 95.9%

#9 · Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-fable-5-1

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 95.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

93.4%89.8% - 97.1%

#10 · Claude Fable 5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-fable-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 92.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

93.2%89.4% - 97%

#11 · GPT-5.5

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 92.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

93.2%90.7% - 95.7%

#12 · kimi-k3-max

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 91.1%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

92.9%90.4% - 95.5%

#13 · grok-4.5

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 91.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

92.9%90.4% - 95.5%

#14 · Gemini 3.5 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.5-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 89.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

92.7%89.8% - 95.5%

#15 · minimax-m3

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 89.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

92.7%89.8% - 95.5%

#16 · Claude Opus 4.8

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-8

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 87.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

92.4%88.5% - 96.3%

#17 · deepseek-v4-pro-0813

VALS-AI · Sep 1, 2026

Source label: deepseek/deepseek-v4-pro-0813

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 87.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: DeepSeek.

92.4%88.5% - 96.4%

#18 · GPT-5.6 Luna

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.6-luna

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

91.7%88.3% - 95.1%

#19 · Gemini 3 Pro Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3-pro-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

91.7%88.9% - 94.4%

#20 · gpt-5.2-2025-12-11

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.2-2025-12-11

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

91.7%88.1% - 95.3%

#21 · GPT-5.4

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.4-2026-03-05

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

91.7%87.9% - 95.4%

#22 · Grok 4.3

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 83.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

91.4%88.7% - 94.2%

#23 · muse-spark-1.1

VALS-AI · Sep 1, 2026

Source label: meta/muse_spark_1_1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 82.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Meta.

91.2%87% - 95.3%

#24 · GPT-5.6 Terra

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.6-terra

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 81.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

90.9%87.2% - 94.6%

#25 · Qwen3.7 Max

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.7-max

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 80.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

90.2%87.2% - 93.1%

#26 · Claude Opus 4.7

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-7

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 80.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

90.2%87.1% - 93.2%

#27 · deepseek-v4-flash-0731

VALS-AI · Sep 1, 2026

Source label: deepseek/deepseek-v4-flash-0731

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 79%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: DeepSeek.

89.9%86.6% - 93.2%

#28 · muse-spark

VALS-AI · Sep 1, 2026

Source label: meta/muse_spark

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 78.2%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Meta.

89.6%86.6% - 92.7%

#29 · Claude Opus 4.6

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-6-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 78.2%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

89.6%87.3% - 92%

#30 · deepseek-v4-pro

VALS-AI · Sep 1, 2026

Source label: deepseek/deepseek-v4-pro

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 76.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: DeepSeek.

89.4%86.2% - 92.6%

#31 · kimi-k2.6

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k2.6

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 75.8%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

89.1%85.2% - 93.1%

#32 · Claude Sonnet 5 (Adaptive Reasoning, Max Effort)

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 75%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

88.9%84.5% - 93.2%

#33 · qwen3.8-27b

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.8-27b

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 75%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

88.9%85.8% - 92%

#34 · Grok 4.20

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.20-0309-reasoning

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 73.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

88.6%85.5% - 91.8%

#35 · Grok 4

VALS-AI · Sep 1, 2026

Source label: grok/grok-4-0709

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 72.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

88.1%84.9% - 91.3%

#36 · glm-5.3-max

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 72.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

88.1%84.5% - 91.8%

#37 · Gemini 3 Flash Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3-flash-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 71%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

87.9%84.7% - 91.1%

#38 · Qwen3.5 Plus

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.5-plus-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 70.2%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

87.4%84.1% - 90.6%

#39 · qwen3.6-plus

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.6-plus

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 70.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

87.4%84.1% - 90.6%

#40 · inkling

VALS-AI · Sep 1, 2026

Source label: thinkingmachines/inkling

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 68.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Thinkingmachines.

87.1%83.6% - 90.6%

#41 · GPT-5.1

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.1-2025-11-13

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 67.7%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

86.6%83.3% - 90%

#42 · minimax-m2.7

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M2.7

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 67.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

86.6%82.8% - 90.4%

#43 · glm-5.3-flash

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.3-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 66.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

86.4%82.2% - 90.5%

#44 · Nemotron 3 Ultra 550B A55B (Reasoning)

VALS-AI · Sep 1, 2026

Source label: nvidia/nemotron-3-ultra-550b-a55b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 65.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Nvidia.

86.1%82.7% - 89.5%

#45 · claude-opus-4-5-20251101-thinking

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-5-20251101-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 64.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

85.9%81% - 90.7%

#46 · GPT-5

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5-2025-08-07

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 63.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

85.6%82.1% - 89.1%

#47 · glm-5.2-max

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.2

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 63.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

85.6%82.1% - 89.1%

#48 · Claude Sonnet 4.6

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 63.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

85.6%80.9% - 90.3%

#49 · Grok 4 Fast

VALS-AI · Sep 1, 2026

Source label: grok/grok-4-fast-reasoning

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 61.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

85.4%81.8% - 88.9%

#50 · Ling 3.0 Flash

VALS-AI · Sep 1, 2026

Source label: ant/ling-3.0-flash-2607

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 60.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Ant.

84.8%80.5% - 89.2%

#51 · Qwen3 Max

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3-max-2026-01-23

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 60.5%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

84.8%81.3% - 88.4%

#52 · glm-5.1

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.1

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 58.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

84.5%80.9% - 88.1%

#53 · Grok 4.1 Fast

VALS-AI · Sep 1, 2026

Source label: grok/grok-4-1-fast-reasoning

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 58.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

84.3%80.7% - 88%

#54 · o3

VALS-AI · Sep 1, 2026

Source label: openai/o3-2025-04-16

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 57.3%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

84.1%80.4% - 87.7%

#55 · kimi-k2.5-thinking

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k2.5-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 57.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

84.1%79.9% - 88.3%

#56 · gemini-3.5-flash-lite

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.5-flash-lite

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 55.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

83.8%80.2% - 87.5%

#57 · Inkling Small

VALS-AI · Sep 1, 2026

Source label: thinkingmachines/inkling-small

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 54.8%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Thinkingmachines.

83.6%79.6% - 87.6%

#58 · glm-5

VALS-AI · Sep 1, 2026

Source label: zai/glm-5-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 54%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

83.3%79.7% - 87%

#59 · GPT-5.4 mini

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 53.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

83.1%78.3% - 87.9%

#60 · Qwen3.5 Flash

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.5-flash

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 52.4%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

82.8%79.1% - 86.5%

#61 · mimo-v2.5-pro

VALS-AI · Sep 1, 2026

Source label: xiaomi/mimo-v2.5-pro

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 51.6%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Xiaomi.

82.6%77.7% - 87.4%

#62 · minimax-m2.5

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 50.8%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

82.1%77.6% - 86.5%

#63 · claude-sonnet-4-5-20250929-thinking

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-5-20250929-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 50%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

81.6%77.2% - 86.1%

#64 · Gemini 2.5 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.5-flash-preview-09-2025

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 49.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

81.6%77.7% - 85.5%

#65 · mimo-v2.5

VALS-AI · Sep 1, 2026

Source label: xiaomi/mimo-v2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 49.2%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Xiaomi.

81.6%77.5% - 85.6%

#66 · Gemini 3.1 Flash-Lite Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.1-flash-lite-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 47.6%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

81.1%77.2% - 84.9%

#67 · Gemini 2.5 Pro

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.5-pro-exp-03-25

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 46.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

80.8%76.9% - 84.7%

#68 · gpt-5-mini-2025-08-07

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5-mini-2025-08-07

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 46%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

80.3%76.4% - 84.2%

#69 · deepseek-v3p2-thinking

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3p2-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 46%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

80.3%76.4% - 84.2%

#70 · glm-4.7

VALS-AI · Sep 1, 2026

Source label: zai/glm-4.7

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 44.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

80.1%76.1% - 84%

#71 · Claude Opus 4.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-5-20251101

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 43.5%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

79.5%74.7% - 84.4%

#72 · Grok 3 mini

VALS-AI · Sep 1, 2026

Source label: grok/grok-3-mini-fast-high-reasoning

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 42.7%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

79.3%75.3% - 83.3%

#73 · GPT-OSS 120B

VALS-AI · Sep 1, 2026

Source label: fireworks/gpt-oss-120b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 41.9%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

78.5%74.1% - 83%

#74 · MiniMax-M2.1

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M2.1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 41.9%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

78.5%74.4% - 82.7%

#75 · Kimi K2 Thinking

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k2-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 41.9%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

78.5%74.3% - 82.8%

#76 · GPT-5.4 nano

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.4-nano-2026-03-17

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 39.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

77.5%72.8% - 82.3%

#77 · claude-opus-4-1-20250805-thinking

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-1-20250805-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 38.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

76.3%72% - 80.5%

#78 · deepseek-v3p2

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3p2

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 38.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

76.3%72% - 80.5%

#79 · o3 mini

VALS-AI · Sep 1, 2026

Source label: openai/o3-mini-2025-01-31

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 37.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

75.5%71.2% - 79.8%

#80 · Claude Sonnet 3.7

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-3-7-sonnet-20250219-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 36.3%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

75.3%71% - 79.5%

#81 · Claude Sonnet 4

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-20250514-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 35.5%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

75%70.7% - 79.3%

#82 · Claude Sonnet 4.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-20250514-thinking

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 35.5%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-5-to-4.

75%70.7% - 79.3%

#83 · o4 mini

VALS-AI · Sep 1, 2026

Source label: openai/o4-mini-2025-04-16

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 33.9%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

74.5%70.2% - 78.8%

#84 · glm-4.6

VALS-AI · Sep 1, 2026

Source label: zai/glm-4.6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 33.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

74.5%70.2% - 78.8%

#85 · Grok 3

VALS-AI · Sep 1, 2026

Source label: grok/grok-3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 32.3%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

74.2%69.9% - 78.6%

#86 · o1

VALS-AI · Sep 1, 2026

Source label: openai/o1-2024-12-17

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 31.5%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

73.2%68.9% - 77.6%

#87 · Claude Haiku 4.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-haiku-4-5-20251001-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 30.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

72.2%67.4% - 77.1%

#88 · glm-4.5

VALS-AI · Sep 1, 2026

Source label: zai/glm-4.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 30.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

72.2%67.8% - 76.7%

#89 · Claude Opus 4

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-20250514

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 29%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

71.7%67.3% - 76.1%

#90 · kimi-k2-instruct

VALS-AI · Sep 1, 2026

Source label: together/moonshotai/Kimi-K2-Instruct

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 28.2%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

71.5%67% - 75.9%

#91 · Gemini 2.5 Flash-Lite

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.5-flash-lite-preview-09-2025-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 27.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

70.2%65.7% - 74.7%

#92 · Qwen3 235B A22B

VALS-AI · Sep 1, 2026

Source label: fireworks/qwen3-235b-a22b

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 27.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

70.2%65.6% - 74.9%

#93 · Claude Opus 4.1

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-1-20250805

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 25.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

70%65.4% - 74.5%

#94 · llama4-maverick-instruct-basic

VALS-AI · Sep 1, 2026

Source label: fireworks/llama4-maverick-instruct-basic

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 25%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

69.4%64.8% - 74.1%

#95 · GPT-OSS 20B

VALS-AI · Sep 1, 2026

Source label: fireworks/gpt-oss-20b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 24.2%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

68.9%64.1% - 73.8%

#96 · Mistral Large (Feb '24)

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-large-2512

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 23.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

68.4%63.8% - 73.1%

#97 · GPT-4.1 mini

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4.1-mini-2025-04-14

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 22.6%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

67.9%63.3% - 72.5%

#98 · GPT-4.1

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4.1-2025-04-14

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 21.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

65.4%60.7% - 70.1%

#99 · Gemini 2.0 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.0-flash-001

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 21%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

65.2%60.5% - 69.8%

#100 · llama-3-3-nemotron-super-49b-v1-42e84561-thinking

VALS-AI · Sep 1, 2026

Source label: together/langston/nim/nvidia/llama-3.3-nemotron-super-49b-v1-42e84561-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 20.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

62.4%57.6% - 67.2%

#101 · Magistral Medium 1.2

VALS-AI · Sep 1, 2026

Source label: mistralai/magistral-medium-2509

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 20.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

62.4%57.6% - 67.1%

#102 · deepseek-v3-0324

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3-0324

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 18.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

61.6%56.8% - 66.4%

#103 · Claude Sonnet 3.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-3-5-sonnet-20241022

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 17.7%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

59.3%54.5% - 64.2%

#104 · mimo-v2-flash (non-thinking)

VALS-AI · Sep 1, 2026

Source label: xiaomi/mimo-v2-flash

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 17.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Xiaomi.

59.3%54.5% - 64.2%

#105 · Gemini 1.5 Pro

VALS-AI · Sep 1, 2026

Source label: google/gemini-1.5-pro-002

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 16.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

58.3%53.5% - 63.2%

#106 · Magistral Small 1.2

VALS-AI · Sep 1, 2026

Source label: mistralai/magistral-small-2509

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 16.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

58.3%53.5% - 63.2%

#107 · laguna-xs.2

VALS-AI · Sep 1, 2026

Source label: poolside/laguna-xs.2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 14.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Poolside.

55.1%50.2% - 59.9%

#108 · deepseek-v3

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 13.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

54.5%49.6% - 59.5%

#109 · GPT-4o

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4o-2024-11-20

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

53.8%48.9% - 58.7%

#110 · GPT-4.1 nano

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4.1-nano-2025-04-14

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

50.8%45.9% - 55.6%

#111 · Mistral Small (Feb '24)

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-small-2402

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

50.8%45.9% - 55.6%

#112 · Grok 2

VALS-AI · Sep 1, 2026

Source label: grok/grok-2-1212

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

50.8%45.8% - 55.7%

#113 · llama-3-3-70b-instruct-turbo

VALS-AI · Sep 1, 2026

Source label: together/meta-llama/Llama-3.3-70B-Instruct-Turbo

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 9.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

50%45.1% - 54.9%

#114 · command-a-03-2025

VALS-AI · Sep 1, 2026

Source label: cohere/command-a-03-2025

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 8.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Cohere.

48.5%44% - 52.9%

#115 · mistral-large-2411

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-large-2411

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 8.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

47.7%42.8% - 52.6%

#116 · llama-3-3-nemotron-super-49b-v1-42e84561

VALS-AI · Sep 1, 2026

Source label: together/langston/nim/nvidia/llama-3.3-nemotron-super-49b-v1-42e84561

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 7.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

47.7%42.9% - 52.6%

#117 · llama-4-scout-17b-16e-instruct

VALS-AI · Sep 1, 2026

Source label: together/meta-llama/Llama-4-Scout-17B-16E-Instruct

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 6.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

47%42.1% - 51.9%

#118 · Gemini 1.5 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-1.5-flash-002

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 5.6%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

46%41.1% - 50.9%

#119 · Mistral Small (Sep '24)

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-small-2503

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 4.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

44.2%39.3% - 49%

#120 · GPT-4o mini

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4o-mini-2024-07-18

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 4.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

44.2%39.3% - 49.1%

#121 · Claude Haiku 3.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-3-5-haiku-20241022

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 3.2%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

37.9%33.1% - 42.7%

#122 · mistral-medium-3.5

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-medium-3.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 2.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

34.8%30.2% - 39.5%

#123 · command-r-plus

VALS-AI · Sep 1, 2026

Source label: cohere/command-r-plus

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 1.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Cohere.

31.1%26.6% - 35.5%

#124 · GPT-3.5 Turbo

VALS-AI · Sep 1, 2026

Source label: openai/gpt-3.5-turbo

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 0.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

30.6%26.1% - 35%

#125 · laguna-m.1

VALS-AI · Sep 1, 2026

Source label: poolside/laguna-m.1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 0%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Poolside.

27%23.3% - 30.8%

Benchmarks · /benchmarks/vals-ai-gpqa

GPQA Diamond

GPQA Diamond result as reported by Vals AI.

Source · Vals AI
Version · vals-ai snapshot 2026-10-03
Scores · 125

Test details

Visible tradeoffsThis is an objective signal, so it is mainly about measurable task performance rather than public taste.

source

Vals AI

metric

Accuracy (%)

judge

Objective

direction

higher better

group id

vals_gpqa_current

domain

Reasoning / math / science

What it measures vs what it misses

✓ Measures

Graduate-level Google-proof scientific reasoning.

✗ Misses

Adjacent skills outside the benchmark task mix, latency, and cost.

Leaderboard · this benchmark version

#1 · Gemini 3.1 Pro Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.1-pro-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 100%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

95.5%93.4% - 97.5%

#2 · GPT-5.6 Sol

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.6-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 99.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

95.2%93.1% - 97.3%

#3 · grok-4.6-high

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 98.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

94.7%92.5% - 96.9%

#4 · gemini-3.8-flash-high

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.8-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 97.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

94.4%91.5% - 97.3%

#5 · gemini-3.7-flash-high

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.7-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 96.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

93.9%91% - 96.9%

#6 · qwen3.8-max

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.8-max

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 96%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

93.7%91.3% - 96.1%

#7 · gemini-3.6-flash-high

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.6-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 95.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

93.4%90.8% - 96%

#8 · Claude Opus 5 (Adaptive Reasoning, Max Effort)

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 95.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

93.4%91% - 95.9%

#9 · Claude Fable 5.1 (Adaptive Reasoning, Max Effort, Default Fallback)

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-fable-5-1

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 95.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

93.4%89.8% - 97.1%

#10 · Claude Fable 5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-fable-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 92.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

93.2%89.4% - 97%

#11 · GPT-5.5

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 92.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

93.2%90.7% - 95.7%

#12 · kimi-k3-max

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 91.1%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

92.9%90.4% - 95.5%

#13 · grok-4.5

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 91.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

92.9%90.4% - 95.5%

#14 · Gemini 3.5 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.5-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 89.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

92.7%89.8% - 95.5%

#15 · minimax-m3

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 89.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

92.7%89.8% - 95.5%

#16 · Claude Opus 4.8

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-8

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 87.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

92.4%88.5% - 96.3%

#17 · deepseek-v4-pro-0813

VALS-AI · Sep 1, 2026

Source label: deepseek/deepseek-v4-pro-0813

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 87.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: DeepSeek.

92.4%88.5% - 96.4%

#18 · GPT-5.6 Luna

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.6-luna

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

91.7%88.3% - 95.1%

#19 · Gemini 3 Pro Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3-pro-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

91.7%88.9% - 94.4%

#20 · gpt-5.2-2025-12-11

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.2-2025-12-11

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

91.7%88.1% - 95.3%

#21 · GPT-5.4

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.4-2026-03-05

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 86.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

91.7%87.9% - 95.4%

#22 · Grok 4.3

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 83.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

91.4%88.7% - 94.2%

#23 · muse-spark-1.1

VALS-AI · Sep 1, 2026

Source label: meta/muse_spark_1_1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 82.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Meta.

91.2%87% - 95.3%

#24 · GPT-5.6 Terra

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.6-terra

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 81.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

90.9%87.2% - 94.6%

#25 · Qwen3.7 Max

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.7-max

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 80.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

90.2%87.2% - 93.1%

#26 · Claude Opus 4.7

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-7

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 80.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

90.2%87.1% - 93.2%

#27 · deepseek-v4-flash-0731

VALS-AI · Sep 1, 2026

Source label: deepseek/deepseek-v4-flash-0731

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 79%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: DeepSeek.

89.9%86.6% - 93.2%

#28 · muse-spark

VALS-AI · Sep 1, 2026

Source label: meta/muse_spark

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 78.2%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Meta.

89.6%86.6% - 92.7%

#29 · Claude Opus 4.6

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-6-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 78.2%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

89.6%87.3% - 92%

#30 · deepseek-v4-pro

VALS-AI · Sep 1, 2026

Source label: deepseek/deepseek-v4-pro

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 76.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: DeepSeek.

89.4%86.2% - 92.6%

#31 · kimi-k2.6

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k2.6

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 75.8%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

89.1%85.2% - 93.1%

#32 · Claude Sonnet 5 (Adaptive Reasoning, Max Effort)

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 75%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

88.9%84.5% - 93.2%

#33 · qwen3.8-27b

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.8-27b

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 75%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

88.9%85.8% - 92%

#34 · Grok 4.20

VALS-AI · Sep 1, 2026

Source label: grok/grok-4.20-0309-reasoning

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 73.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

88.6%85.5% - 91.8%

#35 · Grok 4

VALS-AI · Sep 1, 2026

Source label: grok/grok-4-0709

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 72.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

88.1%84.9% - 91.3%

#36 · glm-5.3-max

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 72.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

88.1%84.5% - 91.8%

#37 · Gemini 3 Flash Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3-flash-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 71%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

87.9%84.7% - 91.1%

#38 · Qwen3.5 Plus

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.5-plus-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 70.2%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

87.4%84.1% - 90.6%

#39 · qwen3.6-plus

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.6-plus

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 70.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

87.4%84.1% - 90.6%

#40 · inkling

VALS-AI · Sep 1, 2026

Source label: thinkingmachines/inkling

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 68.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Thinkingmachines.

87.1%83.6% - 90.6%

#41 · GPT-5.1

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.1-2025-11-13

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 67.7%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

86.6%83.3% - 90%

#42 · minimax-m2.7

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M2.7

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 67.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

86.6%82.8% - 90.4%

#43 · glm-5.3-flash

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.3-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 66.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

86.4%82.2% - 90.5%

#44 · Nemotron 3 Ultra 550B A55B (Reasoning)

VALS-AI · Sep 1, 2026

Source label: nvidia/nemotron-3-ultra-550b-a55b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 65.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Nvidia.

86.1%82.7% - 89.5%

#45 · claude-opus-4-5-20251101-thinking

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-5-20251101-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 64.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

85.9%81% - 90.7%

#46 · GPT-5

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5-2025-08-07

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 63.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

85.6%82.1% - 89.1%

#47 · glm-5.2-max

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.2

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 63.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

85.6%82.1% - 89.1%

#48 · Claude Sonnet 4.6

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 63.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

85.6%80.9% - 90.3%

#49 · Grok 4 Fast

VALS-AI · Sep 1, 2026

Source label: grok/grok-4-fast-reasoning

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 61.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

85.4%81.8% - 88.9%

#50 · Ling 3.0 Flash

VALS-AI · Sep 1, 2026

Source label: ant/ling-3.0-flash-2607

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 60.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Ant.

84.8%80.5% - 89.2%

#51 · Qwen3 Max

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3-max-2026-01-23

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 60.5%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

84.8%81.3% - 88.4%

#52 · glm-5.1

VALS-AI · Sep 1, 2026

Source label: zai/glm-5.1

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 58.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

84.5%80.9% - 88.1%

#53 · Grok 4.1 Fast

VALS-AI · Sep 1, 2026

Source label: grok/grok-4-1-fast-reasoning

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 58.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

84.3%80.7% - 88%

#54 · o3

VALS-AI · Sep 1, 2026

Source label: openai/o3-2025-04-16

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 57.3%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

84.1%80.4% - 87.7%

#55 · kimi-k2.5-thinking

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k2.5-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 57.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

84.1%79.9% - 88.3%

#56 · gemini-3.5-flash-lite

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.5-flash-lite

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 55.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

83.8%80.2% - 87.5%

#57 · Inkling Small

VALS-AI · Sep 1, 2026

Source label: thinkingmachines/inkling-small

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 54.8%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Thinkingmachines.

83.6%79.6% - 87.6%

#58 · glm-5

VALS-AI · Sep 1, 2026

Source label: zai/glm-5-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 54%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

83.3%79.7% - 87%

#59 · GPT-5.4 mini

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 53.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

83.1%78.3% - 87.9%

#60 · Qwen3.5 Flash

VALS-AI · Sep 1, 2026

Source label: alibaba/qwen3.5-flash

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 52.4%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Alibaba.

82.8%79.1% - 86.5%

#61 · mimo-v2.5-pro

VALS-AI · Sep 1, 2026

Source label: xiaomi/mimo-v2.5-pro

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 51.6%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Xiaomi.

82.6%77.7% - 87.4%

#62 · minimax-m2.5

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 50.8%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

82.1%77.6% - 86.5%

#63 · claude-sonnet-4-5-20250929-thinking

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-5-20250929-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 50%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

81.6%77.2% - 86.1%

#64 · Gemini 2.5 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.5-flash-preview-09-2025

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 49.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

81.6%77.7% - 85.5%

#65 · mimo-v2.5

VALS-AI · Sep 1, 2026

Source label: xiaomi/mimo-v2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 49.2%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Xiaomi.

81.6%77.5% - 85.6%

#66 · Gemini 3.1 Flash-Lite Preview

VALS-AI · Sep 1, 2026

Source label: google/gemini-3.1-flash-lite-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 47.6%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

81.1%77.2% - 84.9%

#67 · Gemini 2.5 Pro

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.5-pro-exp-03-25

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 46.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

80.8%76.9% - 84.7%

#68 · gpt-5-mini-2025-08-07

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5-mini-2025-08-07

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 46%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

80.3%76.4% - 84.2%

#69 · deepseek-v3p2-thinking

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3p2-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 46%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

80.3%76.4% - 84.2%

#70 · glm-4.7

VALS-AI · Sep 1, 2026

Source label: zai/glm-4.7

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 44.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

80.1%76.1% - 84%

#71 · Claude Opus 4.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-5-20251101

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 43.5%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

79.5%74.7% - 84.4%

#72 · Grok 3 mini

VALS-AI · Sep 1, 2026

Source label: grok/grok-3-mini-fast-high-reasoning

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 42.7%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

79.3%75.3% - 83.3%

#73 · GPT-OSS 120B

VALS-AI · Sep 1, 2026

Source label: fireworks/gpt-oss-120b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 41.9%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

78.5%74.1% - 83%

#74 · MiniMax-M2.1

VALS-AI · Sep 1, 2026

Source label: minimax/MiniMax-M2.1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 41.9%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: MiniMax.

78.5%74.4% - 82.7%

#75 · Kimi K2 Thinking

VALS-AI · Sep 1, 2026

Source label: kimi/kimi-k2-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 41.9%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Moonshot AI.

78.5%74.3% - 82.8%

#76 · GPT-5.4 nano

VALS-AI · Sep 1, 2026

Source label: openai/gpt-5.4-nano-2026-03-17

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 39.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

77.5%72.8% - 82.3%

#77 · claude-opus-4-1-20250805-thinking

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-1-20250805-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 38.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

76.3%72% - 80.5%

#78 · deepseek-v3p2

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3p2

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 38.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

76.3%72% - 80.5%

#79 · o3 mini

VALS-AI · Sep 1, 2026

Source label: openai/o3-mini-2025-01-31

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 37.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

75.5%71.2% - 79.8%

#80 · Claude Sonnet 3.7

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-3-7-sonnet-20250219-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 36.3%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

75.3%71% - 79.5%

#81 · Claude Sonnet 4

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-20250514-thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 35.5%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

75%70.7% - 79.3%

#82 · Claude Sonnet 4.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-sonnet-4-20250514-thinking

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 35.5%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-5-to-4.

75%70.7% - 79.3%

#83 · o4 mini

VALS-AI · Sep 1, 2026

Source label: openai/o4-mini-2025-04-16

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 33.9%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

74.5%70.2% - 78.8%

#84 · glm-4.6

VALS-AI · Sep 1, 2026

Source label: zai/glm-4.6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 33.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

74.5%70.2% - 78.8%

#85 · Grok 3

VALS-AI · Sep 1, 2026

Source label: grok/grok-3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 32.3%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

74.2%69.9% - 78.6%

#86 · o1

VALS-AI · Sep 1, 2026

Source label: openai/o1-2024-12-17

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 31.5%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

73.2%68.9% - 77.6%

#87 · Claude Haiku 4.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-haiku-4-5-20251001-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 30.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

72.2%67.4% - 77.1%

#88 · glm-4.5

VALS-AI · Sep 1, 2026

Source label: zai/glm-4.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 30.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Zhipu AI.

72.2%67.8% - 76.7%

#89 · Claude Opus 4

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-20250514

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 29%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

71.7%67.3% - 76.1%

#90 · kimi-k2-instruct

VALS-AI · Sep 1, 2026

Source label: together/moonshotai/Kimi-K2-Instruct

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 28.2%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

71.5%67% - 75.9%

#91 · Gemini 2.5 Flash-Lite

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.5-flash-lite-preview-09-2025-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 27.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

70.2%65.7% - 74.7%

#92 · Qwen3 235B A22B

VALS-AI · Sep 1, 2026

Source label: fireworks/qwen3-235b-a22b

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 27.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

70.2%65.6% - 74.9%

#93 · Claude Opus 4.1

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-opus-4-1-20250805

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 25.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

70%65.4% - 74.5%

#94 · llama4-maverick-instruct-basic

VALS-AI · Sep 1, 2026

Source label: fireworks/llama4-maverick-instruct-basic

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 25%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

69.4%64.8% - 74.1%

#95 · GPT-OSS 20B

VALS-AI · Sep 1, 2026

Source label: fireworks/gpt-oss-20b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 24.2%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

68.9%64.1% - 73.8%

#96 · Mistral Large (Feb '24)

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-large-2512

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 23.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

68.4%63.8% - 73.1%

#97 · GPT-4.1 mini

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4.1-mini-2025-04-14

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 22.6%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

67.9%63.3% - 72.5%

#98 · GPT-4.1

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4.1-2025-04-14

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 21.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

65.4%60.7% - 70.1%

#99 · Gemini 2.0 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-2.0-flash-001

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 21%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

65.2%60.5% - 69.8%

#100 · llama-3-3-nemotron-super-49b-v1-42e84561-thinking

VALS-AI · Sep 1, 2026

Source label: together/langston/nim/nvidia/llama-3.3-nemotron-super-49b-v1-42e84561-thinking

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 20.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

62.4%57.6% - 67.2%

#101 · Magistral Medium 1.2

VALS-AI · Sep 1, 2026

Source label: mistralai/magistral-medium-2509

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 20.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

62.4%57.6% - 67.1%

#102 · deepseek-v3-0324

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3-0324

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 18.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

61.6%56.8% - 66.4%

#103 · Claude Sonnet 3.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-3-5-sonnet-20241022

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 17.7%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

59.3%54.5% - 64.2%

#104 · mimo-v2-flash (non-thinking)

VALS-AI · Sep 1, 2026

Source label: xiaomi/mimo-v2-flash

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 17.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Xiaomi.

59.3%54.5% - 64.2%

#105 · Gemini 1.5 Pro

VALS-AI · Sep 1, 2026

Source label: google/gemini-1.5-pro-002

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 16.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

58.3%53.5% - 63.2%

#106 · Magistral Small 1.2

VALS-AI · Sep 1, 2026

Source label: mistralai/magistral-small-2509

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 16.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

58.3%53.5% - 63.2%

#107 · laguna-xs.2

VALS-AI · Sep 1, 2026

Source label: poolside/laguna-xs.2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 14.5%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Poolside.

55.1%50.2% - 59.9%

#108 · deepseek-v3

VALS-AI · Sep 1, 2026

Source label: fireworks/deepseek-v3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 13.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Fireworks AI.

54.5%49.6% - 59.5%

#109 · GPT-4o

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4o-2024-11-20

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

53.8%48.9% - 58.7%

#110 · GPT-4.1 nano

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4.1-nano-2025-04-14

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

50.8%45.9% - 55.6%

#111 · Mistral Small (Feb '24)

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-small-2402

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

50.8%45.9% - 55.6%

#112 · Grok 2

VALS-AI · Sep 1, 2026

Source label: grok/grok-2-1212

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 12.1%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: SpaceXAI.

50.8%45.8% - 55.7%

#113 · llama-3-3-70b-instruct-turbo

VALS-AI · Sep 1, 2026

Source label: together/meta-llama/Llama-3.3-70B-Instruct-Turbo

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 9.7%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

50%45.1% - 54.9%

#114 · command-a-03-2025

VALS-AI · Sep 1, 2026

Source label: cohere/command-a-03-2025

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 8.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Cohere.

48.5%44% - 52.9%

#115 · mistral-large-2411

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-large-2411

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 8.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

47.7%42.8% - 52.6%

#116 · llama-3-3-nemotron-super-49b-v1-42e84561

VALS-AI · Sep 1, 2026

Source label: together/langston/nim/nvidia/llama-3.3-nemotron-super-49b-v1-42e84561

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 7.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

47.7%42.9% - 52.6%

#117 · llama-4-scout-17b-16e-instruct

VALS-AI · Sep 1, 2026

Source label: together/meta-llama/Llama-4-Scout-17B-16E-Instruct

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 6.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Together AI.

47%42.1% - 51.9%

#118 · Gemini 1.5 Flash

VALS-AI · Sep 1, 2026

Source label: google/gemini-1.5-flash-002

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 5.6%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Google.

46%41.1% - 50.9%

#119 · Mistral Small (Sep '24)

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-small-2503

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 4.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

44.2%39.3% - 49%

#120 · GPT-4o mini

VALS-AI · Sep 1, 2026

Source label: openai/gpt-4o-mini-2024-07-18

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 4.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

44.2%39.3% - 49.1%

#121 · Claude Haiku 3.5

VALS-AI · Sep 1, 2026

Source label: anthropic/claude-3-5-haiku-20241022

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 3.2%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Anthropic.

37.9%33.1% - 42.7%

#122 · mistral-medium-3.5

VALS-AI · Sep 1, 2026

Source label: mistralai/mistral-medium-3.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 2.4%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Mistral AI.

34.8%30.2% - 39.5%

#123 · command-r-plus

VALS-AI · Sep 1, 2026

Source label: cohere/command-r-plus

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 1.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Cohere.

31.1%26.6% - 35.5%

#124 · GPT-3.5 Turbo

VALS-AI · Sep 1, 2026

Source label: openai/gpt-3.5-turbo

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 0.8%
Last updated: recent
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

30.6%26.1% - 35%

#125 · laguna-m.1

VALS-AI · Sep 1, 2026

Source label: poolside/laguna-m.1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/gpqa
Percentile: 0%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: Poolside.

27%23.3% - 30.8%

GPQA Diamond

Test details

What it measures vs what it misses

✓ Measures

✗ Misses

Leaderboard · this benchmark version

Loading benchmark evidence.

GPQA Diamond

Test details

What it measures vs what it misses

✓ Measures

✗ Misses

Leaderboard · this benchmark version