Benchmarks · /benchmarks/vals-ai-proof_bench

ProofBench

Name: ProofBench
Creator: Vals AI

ProofBench result as reported by Vals AI.

Source · Vals AI
Version · vals-ai snapshot 2026-10-03
Scores · 43

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_proof_bench_current

domain

Reasoning / math / science

What it measures vs what it misses

✓ Measures

Writing math proofs that can be formally verified.

✗ 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 · Claude Sonnet 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

100%100% - 100%

#2 · Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

100%100% - 100%

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

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

100%100% - 100%

#4 · gpt-6.1-sol-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6.1-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

#5 · gpt-6-astra-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6-astra

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

#6 · gemini-4-argon-high

VALS-AI · Oct 1, 2026

Source label: google/gemini-4-argon

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

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

VALS-AI · Oct 1, 2026

Source label: anthropic/claude-opus-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

#8 · Claude Fable 5

VALS-AI · Oct 1, 2026

Source label: anthropic/claude-fable-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 83.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

95%90.7% - 99.3%

#9 · kimi-k3-max

VALS-AI · Oct 1, 2026

Source label: kimi/kimi-k3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 81%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

87%80.4% - 93.6%

#10 · gpt-6-sol-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 78.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

83%75.6% - 90.4%

#11 · GPT-5.6 Sol

VALS-AI · Oct 1, 2026

Source label: openai/gpt-5.6-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 78.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

83%75.6% - 90.4%

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

VALS-AI · Oct 1, 2026

Source label: anthropic/claude-sonnet-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 73.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

77%68.7% - 85.3%

#13 · hy4-preview

VALS-AI · Oct 1, 2026

Source label: tencent/hy4-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 71.4%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Tencent.

75%66.5% - 83.5%

#14 · GPT-5.6 Terra

VALS-AI · Oct 1, 2026

Source label: openai/gpt-5.6-terra

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 69%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

74%65.4% - 82.6%

#15 · mimo-v2.6-pro

VALS-AI · Oct 1, 2026

Source label: xiaomi/mimo-v2.6-pro

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 66.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

70%61% - 79%

#16 · gpt-6-luna-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6-luna

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 64.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

64%54.5% - 73.5%

#17 · mimo-v2.6-flash

VALS-AI · Oct 1, 2026

Source label: xiaomi/mimo-v2.6-flash

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 61.9%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

63%53.5% - 72.5%

#18 · GPT-5.6 Luna

VALS-AI · Oct 1, 2026

Source label: openai/gpt-5.6-luna

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 59.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

60%50.3% - 69.7%

#19 · gemini-3.7-flash-high

VALS-AI · Oct 1, 2026

Source label: google/gemini-3.7-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 57.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

58%48.3% - 67.7%

#20 · muse-spark-1.3-max

VALS-AI · Oct 1, 2026

Source label: meta/muse_spark_1_3_max

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 57.1%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

58%48.3% - 67.7%

#21 · qwen3.8-max

VALS-AI · Oct 1, 2026

Source label: alibaba/qwen3.8-max

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 57.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

58%48.3% - 67.7%

#22 · deepseek-v4-flash-0731

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

56%46.2% - 65.8%

#23 · deepseek-v4.1-flash-max

VALS-AI · Oct 1, 2026

Source label: deepseek/deepseek-v4.1-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

54%44.2% - 63.8%

#24 · grok-4.6-high

VALS-AI · Oct 1, 2026

Source label: grok/grok-4.6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 45.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

51%41.2% - 60.8%

#25 · deepseek-v4-pro-0813

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 42.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

50%40.2% - 59.8%

#26 · glm-5.3-max

VALS-AI · Oct 1, 2026

Source label: zai/glm-5.3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 40.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

49%39.2% - 58.8%

#27 · gemini-3.8-flash-high

VALS-AI · Oct 1, 2026

Source label: google/gemini-3.8-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 38.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

48%38.2% - 57.8%

#28 · muse-spark-1.2 (xHigh)

VALS-AI · Oct 1, 2026

Source label: meta/muse_spark_1_2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 35.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

43%33.2% - 52.8%

#29 · Step 5 Preview

VALS-AI · Oct 1, 2026

Source label: stepfun/step-5-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 33.3%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Stepfun.

42%32.3% - 51.7%

#30 · Gemini 3.5 Flash

VALS-AI · Oct 1, 2026

Source label: google/gemini-3.5-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 31%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

31%21.9% - 40.1%

#31 · grok-4.5

VALS-AI · Oct 1, 2026

Source label: grok/grok-4.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 31%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

31%21.9% - 40.1%

#32 · grok-4.7-xhigh

VALS-AI · Oct 1, 2026

Source label: grok/grok-4.7

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 26.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

26%17.4% - 34.6%

#33 · Gemini 3.1 Pro Preview

VALS-AI · Oct 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/proof_bench
Percentile: 26.2%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

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

26%17.4% - 34.6%

#34 · mimo-v2.5-pro

VALS-AI · Oct 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/proof_bench
Percentile: 21.4%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

22%13.8% - 30.2%

#35 · glm-5.3-flash

VALS-AI · Oct 1, 2026

Source label: zai/glm-5.3-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 19%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

21%13% - 29%

#36 · minimax-m3

VALS-AI · Oct 1, 2026

Source label: minimax/MiniMax-M3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 16.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

18%10.4% - 25.6%

#37 · deepseek-v4-pro

VALS-AI · Oct 1, 2026

Source label: deepseek/deepseek-v4-pro

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 14.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

16%8.8% - 23.2%

#38 · mimo-v2.5

VALS-AI · Oct 1, 2026

Source label: xiaomi/mimo-v2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 14.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

16%8.8% - 23.2%

#39 · qwen3.8-27b

VALS-AI · Oct 1, 2026

Source label: alibaba/qwen3.8-27b

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 14.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

16%8.8% - 23.2%

#40 · mistral-medium-3.5

VALS-AI · Oct 1, 2026

Source label: mistralai/mistral-medium-3.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 7.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

9%3.4% - 14.6%

#41 · Inkling Small

VALS-AI · Oct 1, 2026

Source label: thinkingmachines/inkling-small

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

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

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

6%1.3% - 10.7%

#42 · Mercury 2.5

VALS-AI · Oct 1, 2026

Source label: inception/mercury-2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

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

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Inception.

3%0% - 6.4%

#43 · inkling

VALS-AI · Oct 1, 2026

Source label: thinkingmachines/inkling

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

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

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

0%0% - 0%

Benchmarks · /benchmarks/vals-ai-proof_bench

ProofBench

ProofBench result as reported by Vals AI.

Source · Vals AI
Version · vals-ai snapshot 2026-10-03
Scores · 43

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_proof_bench_current

domain

Reasoning / math / science

What it measures vs what it misses

✓ Measures

Writing math proofs that can be formally verified.

✗ Misses

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

Leaderboard · this benchmark version

#1 · Claude Sonnet 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

100%100% - 100%

#2 · Claude Opus 5.5 (Adaptive Reasoning, Max Effort, Default Fallback)

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

100%100% - 100%

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

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

100%100% - 100%

#4 · gpt-6.1-sol-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6.1-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

#5 · gpt-6-astra-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6-astra

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

#6 · gemini-4-argon-high

VALS-AI · Oct 1, 2026

Source label: google/gemini-4-argon

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

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

VALS-AI · Oct 1, 2026

Source label: anthropic/claude-opus-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 92.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

99%97% - 101%

#8 · Claude Fable 5

VALS-AI · Oct 1, 2026

Source label: anthropic/claude-fable-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 83.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

95%90.7% - 99.3%

#9 · kimi-k3-max

VALS-AI · Oct 1, 2026

Source label: kimi/kimi-k3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 81%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

87%80.4% - 93.6%

#10 · gpt-6-sol-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 78.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

83%75.6% - 90.4%

#11 · GPT-5.6 Sol

VALS-AI · Oct 1, 2026

Source label: openai/gpt-5.6-sol

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 78.6%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

83%75.6% - 90.4%

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

VALS-AI · Oct 1, 2026

Source label: anthropic/claude-sonnet-5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 73.8%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

77%68.7% - 85.3%

#13 · hy4-preview

VALS-AI · Oct 1, 2026

Source label: tencent/hy4-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 71.4%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Tencent.

75%66.5% - 83.5%

#14 · GPT-5.6 Terra

VALS-AI · Oct 1, 2026

Source label: openai/gpt-5.6-terra

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 69%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

74%65.4% - 82.6%

#15 · mimo-v2.6-pro

VALS-AI · Oct 1, 2026

Source label: xiaomi/mimo-v2.6-pro

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 66.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

70%61% - 79%

#16 · gpt-6-luna-max

VALS-AI · Oct 1, 2026

Source label: openai/gpt-6-luna

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 64.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

64%54.5% - 73.5%

#17 · mimo-v2.6-flash

VALS-AI · Oct 1, 2026

Source label: xiaomi/mimo-v2.6-flash

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 61.9%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

63%53.5% - 72.5%

#18 · GPT-5.6 Luna

VALS-AI · Oct 1, 2026

Source label: openai/gpt-5.6-luna

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 59.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

60%50.3% - 69.7%

#19 · gemini-3.7-flash-high

VALS-AI · Oct 1, 2026

Source label: google/gemini-3.7-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 57.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

58%48.3% - 67.7%

#20 · muse-spark-1.3-max

VALS-AI · Oct 1, 2026

Source label: meta/muse_spark_1_3_max

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 57.1%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

58%48.3% - 67.7%

#21 · qwen3.8-max

VALS-AI · Oct 1, 2026

Source label: alibaba/qwen3.8-max

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 57.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

58%48.3% - 67.7%

#22 · deepseek-v4-flash-0731

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

56%46.2% - 65.8%

#23 · deepseek-v4.1-flash-max

VALS-AI · Oct 1, 2026

Source label: deepseek/deepseek-v4.1-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

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

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

54%44.2% - 63.8%

#24 · grok-4.6-high

VALS-AI · Oct 1, 2026

Source label: grok/grok-4.6

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 45.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

51%41.2% - 60.8%

#25 · deepseek-v4-pro-0813

VALS-AI · Oct 1, 2026

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

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 42.9%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

50%40.2% - 59.8%

#26 · glm-5.3-max

VALS-AI · Oct 1, 2026

Source label: zai/glm-5.3

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 40.5%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

49%39.2% - 58.8%

#27 · gemini-3.8-flash-high

VALS-AI · Oct 1, 2026

Source label: google/gemini-3.8-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 38.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

48%38.2% - 57.8%

#28 · muse-spark-1.2 (xHigh)

VALS-AI · Oct 1, 2026

Source label: meta/muse_spark_1_2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 35.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

43%33.2% - 52.8%

#29 · Step 5 Preview

VALS-AI · Oct 1, 2026

Source label: stepfun/step-5-preview

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 33.3%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Stepfun.

42%32.3% - 51.7%

#30 · Gemini 3.5 Flash

VALS-AI · Oct 1, 2026

Source label: google/gemini-3.5-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 31%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

31%21.9% - 40.1%

#31 · grok-4.5

VALS-AI · Oct 1, 2026

Source label: grok/grok-4.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 31%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

31%21.9% - 40.1%

#32 · grok-4.7-xhigh

VALS-AI · Oct 1, 2026

Source label: grok/grok-4.7

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 26.2%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

26%17.4% - 34.6%

#33 · Gemini 3.1 Pro Preview

VALS-AI · Oct 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/proof_bench
Percentile: 26.2%
Last updated: recent
Eligibility: preview_model
Identity: provider alias (0.92)

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

26%17.4% - 34.6%

#34 · mimo-v2.5-pro

VALS-AI · Oct 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/proof_bench
Percentile: 21.4%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

22%13.8% - 30.2%

#35 · glm-5.3-flash

VALS-AI · Oct 1, 2026

Source label: zai/glm-5.3-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 19%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

21%13% - 29%

#36 · minimax-m3

VALS-AI · Oct 1, 2026

Source label: minimax/MiniMax-M3

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 16.7%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

18%10.4% - 25.6%

#37 · deepseek-v4-pro

VALS-AI · Oct 1, 2026

Source label: deepseek/deepseek-v4-pro

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 14.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

16%8.8% - 23.2%

#38 · mimo-v2.5

VALS-AI · Oct 1, 2026

Source label: xiaomi/mimo-v2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 14.3%
Last updated: recent
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

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

16%8.8% - 23.2%

#39 · qwen3.8-27b

VALS-AI · Oct 1, 2026

Source label: alibaba/qwen3.8-27b

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 14.3%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

16%8.8% - 23.2%

#40 · mistral-medium-3.5

VALS-AI · Oct 1, 2026

Source label: mistralai/mistral-medium-3.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.vals.ai/benchmarks/proof_bench
Percentile: 7.1%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

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

9%3.4% - 14.6%

#41 · Inkling Small

VALS-AI · Oct 1, 2026

Source label: thinkingmachines/inkling-small

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

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

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

6%1.3% - 10.7%

#42 · Mercury 2.5

VALS-AI · Oct 1, 2026

Source label: inception/mercury-2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

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

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Inception.

3%0% - 6.4%

#43 · inkling

VALS-AI · Oct 1, 2026

Source label: thinkingmachines/inkling

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

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

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

0%0% - 0%

ProofBench

Test details

What it measures vs what it misses

✓ Measures

✗ Misses

Leaderboard · this benchmark version

Loading benchmark evidence.

ProofBench

Test details

What it measures vs what it misses

✓ Measures

✗ Misses

Leaderboard · this benchmark version