Benchmarks · /benchmarks/terminal-bench-2

Terminal-Bench 2.0

Name: Terminal-Bench 2.0
Creator: Terminal-Bench

Official Terminal-Bench 2.0 leaderboard for realistic multi-step terminal tasks.

Source · Terminal-Bench
Version · terminal-bench snapshot 2026-06-24
Scores · 31

Test details

Verified but agingThis is an objective signal, so it is mainly about measurable task performance rather than public taste.

source

Terminal-Bench

metric

Accuracy (%)

judge

Objective

direction

higher better

group id

terminal_bench_2_live

domain

Coding

What it measures vs what it misses

✓ Measures

End-to-end task success on hard terminal workflows that require planning, editing, debugging, and execution.

✗ Misses

IDE-native workflows, code review quality, and non-terminal product engineering work.

Why this countsIt tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.Same-test ruleThis percentile only compares models inside the exact benchmark/version group shown here. It is not a universal score.What it missesIt does not fully capture repo-scale iteration, IDE ergonomics, or long debugging loops.

Leaderboard · this benchmark version

#1 · GPT-5.5

TERMINAL-BENCH · Apr 23, 2026

Source label: gpt-5.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 100%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.121.0). Display model: GPT-5.5. Integration method: API. Agent URL: https://developers.openai.com/codex/cli. Reported stderr: 1.135 percentage points.

82.2%

#2 · GPT-5.3 Codex

TERMINAL-BENCH · Feb 6, 2026

Source label: gpt-5.3-codex

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 96.7%
Last updated: aging
Eligibility: specialized_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Simple Codex (unknown). Display model: GPT-5.3-Codex. Integration method: API. Agent URL: https://openai.com/codex. Reported stderr: 1.221 percentage points.

75.1%

#3 · Gemini 3 Pro Preview

TERMINAL-BENCH · Jan 6, 2026

verified runtimeexact directBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 93.3%
Last updated: archived
Eligibility: preview_model
Identity: exact (1.00)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Ante (unknown). Display model: Gemini 3 Pro. Integration method: API. Agent URL: https://antigma.ai. Reported stderr: 1.066 percentage points.

69.4%

#4 · Claude Opus 4.6

TERMINAL-BENCH · Feb 6, 2026

verified runtimeexact directBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 90%
Last updated: aging
Eligibility: historical_model
Identity: exact (1.00)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.6. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.358 percentage points.

62.9%

#5 · GPT-5.2

TERMINAL-BENCH · Dec 18, 2025

Source label: gpt-5.2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 90%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.73.0). Display model: GPT-5.2. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.532 percentage points.

62.9%

#6 · GPT-5.1

TERMINAL-BENCH · Nov 24, 2025

Source label: gpt-5.1-codex-max

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 83.3%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.63.0). Display model: GPT-5.1-Codex-Max. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.358 percentage points.

60.4%

#7 · Claude Opus 4.5

TERMINAL-BENCH · Nov 22, 2025

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

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 80%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.5. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.291 percentage points.

57.8%

#8 · Gemini 3 Flash Preview

TERMINAL-BENCH · Jan 7, 2026

verified runtimeexact directBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 76.7%
Last updated: archived
Eligibility: preview_model
Identity: exact (1.00)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Gemini 3 Flash. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.597 percentage points.

51.7%

#9 · GPT-5

TERMINAL-BENCH · Nov 4, 2025

verified runtimeexact direct

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 73.3%
Last updated: archived
Eligibility: headline eligible
Identity: exact (1.00)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.53.0). Display model: GPT-5. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.478 percentage points.

49.6%

#10 · GPT-5.4

TERMINAL-BENCH · Nov 4, 2025

Source label: gpt-5

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 73.3%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

49.6%

#11 · Kimi K2.5 (Reasoning)

TERMINAL-BENCH · Feb 4, 2026

Source label: kimi-k2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 66.7%
Last updated: aging
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Kimi K2.5. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.468 percentage points.

43.2%

#12 · Claude Sonnet 4.5

TERMINAL-BENCH · Oct 31, 2025

Source label: claude-sonnet-4-5-20250929

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 63.3%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Sonnet 4.5. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.423 percentage points.

42.8%

#13 · Claude Opus 4.1

TERMINAL-BENCH · Oct 31, 2025

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

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 60%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.1. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.339 percentage points.

38%

#14 · Claude Opus 4

TERMINAL-BENCH · Oct 31, 2025

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

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 60%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.1. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.339 percentage points. Backfilled from Claude Opus 4.1 via approved benchmark identity mapping map-claude-opus-4-to-4-1.

38%

#15 · Claude Opus 4.7

TERMINAL-BENCH · Oct 31, 2025

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

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 60%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.1. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.339 percentage points. Backfilled from Claude Opus 4.1 via approved benchmark identity mapping map-claude-opus-4-7-to-4-1.

38%

#16 · MiniMax-M2.1

TERMINAL-BENCH · Dec 22, 2025

Source label: minimax-m2.1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 50%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Crux (unknown). Display model: MiniMax M2.1. Integration method: API. Agent URL: https://github.com/0xDarkMatter/crux-agent. Reported stderr: 1.474 percentage points.

36.6%

#17 · Kimi K2 Thinking

TERMINAL-BENCH · Nov 11, 2025

Source label: moonshotai/Kimi-K2-Thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 46.7%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Kimi K2 Thinking. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.439 percentage points.

35.7%

#18 · glm-4.7

TERMINAL-BENCH · Jan 28, 2026

Source label: glm-4.7

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 43.3%
Last updated: aging
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: GLM 4.7. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.450 percentage points.

33.4%

#19 · Gemini 2.5 Pro

TERMINAL-BENCH · Oct 31, 2025

Source label: gemini-2.5-pro

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 40%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Gemini 2.5 Pro. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.542 percentage points.

32.6%

#20 · GPT-5.4 mini

TERMINAL-BENCH · Nov 4, 2025

Source label: gpt-5-mini

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 36.7%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.53.0). Display model: GPT-5-Mini. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.530 percentage points.

31.9%

#21 · minimax-m2

TERMINAL-BENCH · Nov 1, 2025

Source label: accounts/fireworks/models/minimax-m2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 33.3%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: MiniMax M2. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.398 percentage points.

30%

#22 · Claude Haiku 4.5

TERMINAL-BENCH · Nov 3, 2025

Source label: claude-haiku-4-5-20251001

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 30%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: Claude Haiku 4.5. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 1.282 percentage points.

29.8%

#23 · Kimi K2 0905

TERMINAL-BENCH · Nov 1, 2025

Source label: moonshotai/Kimi-K2-Instruct-0905

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 26.7%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Kimi K2 Instruct. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.287 percentage points.

27.8%

#24 · Grok 4

TERMINAL-BENCH · Nov 2, 2025

Source label: grok-4-0709

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 23.3%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: OpenHands (0.60.0). Display model: Grok 4. Integration method: API. Agent URL: https://github.com/OpenHands/OpenHands. Reported stderr: 1.557 percentage points.

27.2%

#25 · Grok Code Fast

TERMINAL-BENCH · Nov 3, 2025

Source label: grok-code-fast-1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 20%
Last updated: archived
Eligibility: specialized_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: Grok Code Fast 1. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 1.339 percentage points.

25.8%

#26 · Qwen3-Coder 480B A35B

TERMINAL-BENCH · Nov 2, 2025

Source label: Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 16.7%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: OpenHands (0.60.0). Display model: Qwen 3 Coder 480B. Integration method: API. Agent URL: https://github.com/OpenHands/OpenHands. Reported stderr: 1.320 percentage points.

25.4%

#27 · glm-4.6

TERMINAL-BENCH · Nov 1, 2025

Source label: accounts/fireworks/models/glm-4p6

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 13.3%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: GLM 4.6. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.247 percentage points.

24.5%

#28 · GPT-OSS 120B

TERMINAL-BENCH · Nov 1, 2025

Source label: openai/gpt-oss-120b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 10%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: GPT-OSS-120B. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.357 percentage points.

18.7%

#29 · Gemini 2.5 Flash

TERMINAL-BENCH · Nov 3, 2025

Source label: gemini-2.5-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 6.7%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: Gemini 2.5 Flash. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 1.271 percentage points.

17.1%

#30 · GPT-5.4 nano

TERMINAL-BENCH · Nov 4, 2025

Source label: gpt-5-nano

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 3.3%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.53.0). Display model: GPT-5-Nano. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.157 percentage points.

11.5%

#31 · GPT-OSS 20B

TERMINAL-BENCH · Nov 3, 2025

Source label: openai/gpt-oss-20b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 0%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: GPT-OSS-20B. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 0.693 percentage points.

3.4%

TerminalBench agent leaderboard

#1 · NexAU-AHE

GPT-5.5 · GPT-5.5 · May 14, 2026

china-qijizhifeng · OpenAI · API · unverified

https://github.com/china-qijizhifeng/agentic-harness-engineering.git

84.7%± 1.1%

#2 · LemonHarness

Gemini 3.1 Pro Preview · Gemini 3.1 Pro Preview · May 14, 2026

LR AILab of Lenovo CTO Org · Google, OpenAI · API · unverified

https://github.com/Open-Lemon/LemonAgent

84.5%± 1.3%

#3 · Capy

GPT-5.5 · GPT-5.5 · May 14, 2026

Capy · OpenAI · API · unverified

https://capy.ai

83.1%± 1.1%

#4 · Codex CLI

GPT-5.5 · GPT-5.5 · Apr 23, 2026

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli

82.2%± 1.1%

#5 · Polaris

Claude Opus 4.7 · Claude Opus 4.7 · May 14, 2026

PolarisOps · Anthropic, Google, OpenAI · API · unverified

https://polarops.ai

82.2%± 1.4%

#6 · WOZCODE

Claude Opus 4.7 · Claude Opus 4.7 · May 14, 2026

WOZCODE · Anthropic · API · unverified

https://github.com/WithWoz/wozcode-plugin

80.2%± 1.1%

#7 · TongAgents

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · Mar 13, 2026

BIGAI · Google · API · unverified

https://www.bigai.ai/

80.2%± 1.3%

#8 · LemonHarness

Gemini 3.1 Pro Preview · Gemini 3.1 Pro Preview · May 14, 2026

LR AILab of Lenovo CTO Org · Google, OpenAI · API · unverified

https://github.com/Open-Lemon/LemonAgent

79.9%± 1.5%

#9 · SageAgent

GPT-5.3-Codex · GPT-5.3 Codex · Mar 13, 2026

OpenSage · OpenAI · API · unverified

https://github.com/opensage-agent

78.4%± 1.1%

#10 · Droid

GPT-5.3-Codex · GPT-5.3 Codex · Feb 24, 2026

Factory · OpenAI · API · unverified

https://factory.ai/

77.3%± 1.1%

#11 · Meta-Harness

Claude Opus 4.6 · Claude Opus 4.6 · May 14, 2026

Stanford IRIS · Anthropic · API · unverified

https://github.com/stanford-iris-lab/meta-harness-tbench2-artifact

76.4%± 1.2%

#12 · CodeBrain-1.5

GPT-5.3-Codex · GPT-5.3 Codex · Feb 10, 2026

Feeling AI · OpenAI · API · unverified

https://feeling-ai.com

75.8%± 1%

#13 · Codelia

GPT-5.3-Codex · GPT-5.3 Codex · May 14, 2026

kousw · OpenAI · API · unverified

https://github.com/kousw/codelia

75.7%± 1.1%

#14 · Capy

Claude Opus 4.6 · Claude Opus 4.6 · Mar 12, 2026

Capy · Anthropic · API · unverified

https://capy.ai

75.3%± 1.2%

#15 · Simple Codex

GPT-5.3-Codex · GPT-5.3 Codex · Feb 6, 2026

OpenAI · OpenAI · API · verified

https://openai.com/codex

75.1%± 1.2%

#16 · Terminus-KIRA

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · Feb 23, 2026

KRAFTON AI · Google · API · unverified

https://www.krafton.ai/en/

74.8%± 1.3%

#17 · Terminus-KIRA

Claude Opus 4.6 · Claude Opus 4.6 · Feb 22, 2026

KRAFTON AI · Anthropic · API · unverified

https://www.krafton.ai/en/

74.7%± 1.3%

#18 · Mux

GPT-5.3-Codex · GPT-5.3 Codex · Mar 6, 2026

Coder · OpenAI · API · unverified

https://github.com/coder/mux

74.6%± 1.3%

#19 · MAYA-V2

Claude 4.6 Opus · Mar 12, 2026

ADYA · Anthropic · API · unverified

https://adya.ai/maya

72.1%± 1.1%

#20 · TongAgents

Claude Opus 4.6 · Claude Opus 4.6 · Feb 22, 2026

Bigai · Anthropic · API · unverified

https://www.bigai.ai/

71.9%± 1.4%

#21 · spoox-o-m

GPT-5.3-Codex · GPT-5.3 Codex · May 15, 2026

TUM · OpenAI · API · unverified

https://github.com/plaume8/spoox

71.5%± 1.3%

#22 · Junie CLI

Claude Opus 4.6 · Claude Opus 4.6 · Mar 7, 2026

JetBrains · Anthropic, Google, Google, OpenAI · API · unverified

https://junie.jetbrains.com

71%± 1.5%

#23 · Droid

Claude Opus 4.6 · Claude Opus 4.6 · Feb 5, 2026

Factory · Anthropic · API · unverified

https://factory.ai/

69.9%± 1.3%

#24 · Ante

Gemini 3 Pro · Gemini 3 Pro Preview · Jan 6, 2026

Antigma Labs · Google · API · verified

https://antigma.ai

69.4%± 1.1%

#25 · IndusAGI Coding Agent

GPT-5.3-Codex · GPT-5.3 Codex · Mar 18, 2026

Varun Israni (SoloVpx) · OpenAI · API · unverified

https://github.com/varun-israni/indusagi-coding-agent

69.1%± 1.2%

#26 · Crux

Claude Opus 4.6 · Claude Opus 4.6 · Feb 23, 2026

Roam · Anthropic · API · unverified

https://github.com/0xDarkMatter/crux-agent

66.9%

#27 · Deep Agents

GPT-5.2-Codex · GPT-5.2 · Feb 12, 2026

LangChain · OpenAI · API · unverified

https://github.com/langchain-ai/deepagents

66.5%± 1.6%

#28 · Mux

Claude Opus 4.6 · Claude Opus 4.6 · Feb 13, 2026

Coder · Anthropic · API · unverified

https://github.com/coder/mux

66.5%± 1.3%

#29 · clnkr

GPT-5.5 · GPT-5.5 · May 14, 2026

clnkr · OpenAI · API · unverified

https://github.com/clnkr-ai/clnkr

66.1%± 1.3%

#30 · SageAgent

Gemini 3 Pro · Gemini 3 Pro Preview · Feb 23, 2026

OpenSage · Google · API · unverified

https://github.com/opensage-agent

65.2%± 1.1%

#31 · Droid

GPT-5.2 · GPT-5.2 · Dec 24, 2025

Factory · OpenAI · API · unverified

https://factory.ai/

64.9%± 1.4%

#32 · Terminus 2

GPT-5.3-Codex · GPT-5.3 Codex · Feb 5, 2026

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

64.7%± 1.4%

#33 · Junie CLI

Gemini 3 Flash · Gemini 3 Flash Preview · Dec 23, 2025

JetBrains · Google · API · unverified

https://junie.jetbrains.com

64.3%± 1.4%

#34 · Droid

Claude Opus 4.5 · Claude Opus 4.5 · Dec 11, 2025

Factory · Anthropic · API · unverified

https://factory.ai/

63.1%± 1.4%

#35 · Codex CLI

GPT-5.2 · GPT-5.2 · Dec 18, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

62.9%± 1.5%

#36 · Terminus 2

Claude Opus 4.6 · Claude Opus 4.6 · Feb 6, 2026

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

62.9%± 1.4%

#37 · CodeBrain-1.5

Gemini 3 Pro · Gemini 3 Pro Preview · Feb 5, 2026

Feeling AI · Google · API · unverified

https://feeling-ai.com

62.2%± 1.3%

#38 · II-Agent

Gemini 3 Pro · Gemini 3 Pro Preview · Dec 23, 2025

Intelligent Internet · Google · API · unverified

https://agent.ii.inc/

61.8%± 1.4%

#39 · hookele

GPT-5.1-Codex-Mini · GPT-5.1 · May 14, 2026

Dmitry Barakhov · OpenAI · API · unverified

https://github.com/sady4850/hookele_coding_agent

61.6%± 1%

#40 · Gemini CLI

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · May 14, 2026

Google · Google · API · unverified

https://github.com/google-gemini/gemini-cli

61.4%± 2.1%

#41 · Warp

Claude Haiku 4.5 · Claude Haiku 4.5 · Dec 12, 2025

Warp · Anthropic, OpenAI · API · unverified

https://www.warp.dev/

61.2%± 1.5%

#42 · Droid

Gemini 3 Pro · Gemini 3 Pro Preview · Dec 24, 2025

Factory · Google · API · unverified

https://factory.ai/

61.1%± 1.4%

#43 · Mux

GPT-5.2 · GPT-5.2 · Jan 17, 2026

Coder · OpenAI · API · unverified

https://github.com/coder/mux

60.7%

#44 · Codex CLI

GPT-5.1-Codex-Max · GPT-5.1 · Nov 24, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

60.4%± 1.4%

#45 · Gemini CLI

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · May 14, 2026

Google · Google · API · unverified

https://github.com/google-gemini/gemini-cli

59.4%± 2.1%

#46 · Letta Code

Claude Opus 4.5 · Claude Opus 4.5 · Dec 17, 2025

Letta · Anthropic · API · unverified

https://www.letta.com/

59.1%± 1.2%

#47 · Warp

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 20, 2025

Warp · Anthropic, Google, OpenAI · API · unverified

https://www.warp.dev/

59.1%± 1.4%

#48 · Abacus AI Desktop

Claude Haiku 4.5 · Claude Haiku 4.5 · Dec 11, 2025

Abacus.AI · Anthropic, Google · API · unverified

https://deepagent-desktop.abacus.ai/

58.4%± 1.4%

#49 · Mux

Claude Opus 4.5 · Claude Opus 4.5 · Jan 17, 2026

Coder · Anthropic · API · unverified

https://github.com/coder/mux

58.4%

#50 · Claude Code

Claude Opus 4.6 · Claude Opus 4.6 · Feb 7, 2026

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

58%± 1.5%

#51 · Crux

GPT-5.1-Codex · GPT-5.1 · Nov 16, 2025

Roam · OpenAI · API · verified

https://github.com/0xDarkMatter/crux-agent

57.8%± 1.5%

#52 · Terminus 2

Claude Opus 4.5 · Claude Opus 4.5 · Nov 22, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

57.8%± 1.3%

#53 · Grok CLI

Grok 4.20 Reasoning · Grok 4.20 · Apr 2, 2026

Superagent · xAI · API · unverified

https://github.com/superagent-ai/grok-cli

57.3%

#54 · Terminus 2

Gemini 3 Pro · Gemini 3 Pro Preview · Nov 21, 2025

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

56.9%± 1.3%

#55 · Letta Code

Gemini 3 Pro · Gemini 3 Pro Preview · Dec 17, 2025

Letta · Google · API · unverified

https://www.letta.com/

56%± 1.5%

#56 · Goose

Claude Opus 4.5 · Claude Opus 4.5 · Dec 11, 2025

Block · Anthropic · API · unverified

https://block.github.io/goose/

54.3%± 1.3%

#57 · Terminus 2

GPT-5.2 · GPT-5.2 · Dec 12, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

54%± 1.5%

#58 · Letta Code

GPT-5.1-Codex · GPT-5.1 · Dec 17, 2025

Letta · OpenAI · API · unverified

https://www.letta.com/

53.5%± 1.4%

#59 · Simplai Agent

Claude Sonnet 4.6 · Claude Sonnet 4.6 · May 14, 2026

SimplAI · Anthropic · API · unverified

https://simplai.ai

53.4%± 1.4%

#60 · Terminus 2

GLM 5 · glm-5 · Feb 23, 2026

Terminal-Bench · Z-AI · API · unverified

https://www.tbench.ai/terminus

52.4%± 1.3%

#61 · Claude Code

Claude Opus 4.5 · Claude Opus 4.5 · Dec 18, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

52.1%± 1.3%

#62 · OpenHands

Claude Opus 4.5 · Claude Opus 4.5 · Jan 4, 2026

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

51.9%± 1.5%

#63 · Terminus 2

Gemini 3 Flash · Gemini 3 Flash Preview · Jan 7, 2026

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

51.7%± 1.6%

#64 · OpenCode

Claude Opus 4.5 · Claude Opus 4.5 · Jan 12, 2026

Anomaly Innovations · Anthropic · API · unverified

https://opencode.ai

51.7%

#65 · Warp

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 11, 2025

Warp · Anthropic, Anthropic, OpenAI · API · unverified

https://www.warp.dev/

50.1%± 1.4%

#66 · Codex CLI

GPT-5 · GPT-5 · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

49.6%± 1.5%

#67 · Terminus 2

GPT-5.1 · GPT-5.1 · Nov 16, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

47.6%± 1.4%

#68 · Gemini CLI

Gemini 3 Flash · Gemini 3 Flash Preview · Mar 6, 2026

Google · Google · API · unverified

https://github.com/google-gemini/gemini-cli

47.4%± 1.5%

#69 · CAMEL-AI

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Dec 24, 2025

CAMEL-AI · Anthropic · API · unverified

https://www.camel-ai.org/

46.5%± 1.2%

#70 · IndusAGI Coding Agent

MiniMax M2.7 · minimax-m2.7 · May 14, 2026

Varun Israni (SoloVpx) · Minimax · API · unverified

https://github.com/varun-israni/indusagi-coding-agent

45.1%

#71 · Codex CLI

GPT-5-Codex · GPT-5.3 Codex · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

44.3%± 1.4%

#72 · OpenHands

GPT-5 · GPT-5 · Nov 2, 2025

OpenHands · OpenAI · API · verified

https://github.com/OpenHands/OpenHands

43.8%± 1.5%

#73 · Terminus 2

GPT-5-Codex · GPT-5.3 Codex · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

43.4%± 1.5%

#74 · Terminus 2

Kimi K2.5 · Kimi K2.5 (Reasoning) · Feb 4, 2026

Terminal-Bench · Kimi · API · verified

https://www.tbench.ai/terminus

43.2%± 1.5%

#75 · Goose

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Dec 11, 2025

Block · Anthropic · API · unverified

https://block.github.io/goose/

43.1%± 1.3%

#76 · Crux

GPT-5.1-Codex-Mini · GPT-5.1 · Nov 17, 2025

Roam · OpenAI · API · verified

https://github.com/0xDarkMatter/crux-agent

43.1%± 1.5%

#77 · Harness Agent

MiniMax M2.7 Highspeed · May 14, 2026

lazyFrogLOL · MiniMax · API · unverified

https://github.com/lazyFrogLOL/Harness_Engineering

42.9%± 1.5%

#78 · Terminus 2

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Oct 31, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

42.8%± 1.4%

#79 · MAYA-V2

Claude 4.5 Sonnet · Claude Sonnet 4.5 · Jan 4, 2026

ADYA · Anthropic · API · unverified

https://adya.ai/maya

42.7%

#80 · cchuter

minimax-m2.5 · minimax-m2.5 · Mar 30, 2026

teamblobfish.com · minimax · API · unverified

https://github.com/cchuter/blobfish

42.7%± 1.4%

#81 · OpenHands

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Nov 2, 2025

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

42.6%± 1.4%

#82 · Mini-SWE-Agent

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Nov 3, 2025

Princeton · Anthropic · API · verified

https://github.com/SWE-agent/mini-swe-agent

42.5%± 1.4%

#83 · Terminus 2

Minimax m2.5 · minimax-m2.5 · Feb 23, 2026

Terminal-Bench · Minimax · API · unverified

https://www.tbench.ai/terminus

42.2%± 1.3%

#84 · Mini-SWE-Agent

GPT-5-Codex · GPT-5.3 Codex · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

41.3%± 1.4%

#85 · Claude Code

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Nov 4, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

40.1%± 1.5%

#86 · Terminus 2

DeepSeek-V3.2 · DeepSeek Chat · Feb 10, 2026

Terminal-Bench · DeepSeek · API · unverified

https://www.tbench.ai/terminus

39.6%± 1.4%

#87 · Terminus 2

Claude Opus 4.1 · Claude Opus 4.1 · Oct 31, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

38%± 1.3%

#88 · OpenHands

Claude Opus 4.1 · Claude Opus 4.1 · Nov 2, 2025

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

36.9%± 1.4%

#89 · Terminus 2

GPT-5.1-Codex · GPT-5.1 · Nov 17, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

36.9%± 1.7%

#90 · Crux

MiniMax M2.1 · MiniMax-M2.1 · Dec 22, 2025

Roam · MiniMax · API · verified

https://github.com/0xDarkMatter/crux-agent

36.6%± 1.5%

#91 · Terminus 2

Kimi K2 Thinking · Kimi K2 Thinking · Nov 11, 2025

Terminal-Bench · Moonshot AI · API · verified

https://www.tbench.ai/terminus

35.7%± 1.4%

#92 · Goose

Claude Haiku 4.5 · Claude Haiku 4.5 · Dec 11, 2025

Block · Anthropic · API · unverified

https://block.github.io/goose/

35.5%± 1.5%

#93 · Terminus 2

GPT-5 · GPT-5 · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

35.2%± 1.6%

#94 · Mini-SWE-Agent

Claude Opus 4.1 · Claude Opus 4.1 · Nov 3, 2025

Princeton · Anthropic · API · verified

https://github.com/SWE-agent/mini-swe-agent

35.1%± 1.3%

#95 · Claude Code

Claude Opus 4.1 · Claude Opus 4.1 · Nov 4, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

34.8%± 1.5%

#96 · spoox-o-m

GPT-5-Mini · GPT-5.4 mini · Dec 24, 2025

TUM · OpenAI · API · unverified

https://github.com/plaume8/spoox

34.8%± 1.4%

#97 · Mini-SWE-Agent

GPT-5 · GPT-5 · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

33.9%± 1.5%

#98 · Terminus 2

GLM 4.7 · glm-4.7 · Jan 28, 2026

Terminal-Bench · Z-AI · API · verified

https://www.tbench.ai/terminus

33.4%± 1.4%

#99 · Crux

GLM 4.7 · glm-4.7 · Feb 8, 2026

Roam · Z-AI · API · unverified

https://github.com/0xDarkMatter/crux-agent

33.3%± 1.3%

#100 · Terminus 2

Gemini 2.5 Pro · Gemini 2.5 Pro · Oct 31, 2025

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

32.6%± 1.5%

#101 · Codex CLI

GPT-5-Mini · GPT-5.4 mini · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

31.9%± 1.5%

#102 · Terminus 2

MiniMax M2 · minimax-m2 · Nov 1, 2025

Terminal-Bench · MiniMax · API · verified

https://www.tbench.ai/terminus

30%± 1.4%

#103 · Mini-SWE-Agent

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 3, 2025

Princeton · Anthropic · API · verified

https://github.com/SWE-agent/mini-swe-agent

29.8%± 1.3%

#104 · Terminus 2

MiniMax M2.1 · MiniMax-M2.1 · Dec 23, 2025

Terminal-Bench · MiniMax · API · verified

https://www.tbench.ai/terminus

29.2%± 1.5%

#105 · OpenHands

GPT-5-Mini · GPT-5.4 mini · Nov 2, 2025

OpenHands · OpenAI · API · verified

https://github.com/OpenHands/OpenHands

29.2%± 1.4%

#106 · Terminus 2

Claude Haiku 4.5 · Claude Haiku 4.5 · Oct 31, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

28.3%± 1.5%

#107 · Terminus 2

Kimi K2 Instruct · Kimi K2 0905 · Nov 1, 2025

Terminal-Bench · Moonshot AI · API · verified

https://www.tbench.ai/terminus

27.8%± 1.3%

#108 · Claude Code

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 4, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

27.5%± 1.4%

#109 · OpenHands

Grok 4 · Grok 4 · Nov 2, 2025

OpenHands · xAI · API · verified

https://github.com/OpenHands/OpenHands

27.2%± 1.6%

#110 · Dakou Agent

Qwen 3 Coder 480B · Qwen3-Coder 480B A35B · Dec 28, 2025

iflow · Alibaba · API · unverified

https://dakou.iflow.cn/

27.2%± 1.3%

#111 · OpenHands

Kimi K2 Instruct · Kimi K2 0905 · Nov 2, 2025

OpenHands · Moonshot AI · API · verified

https://github.com/OpenHands/OpenHands

26.7%± 1.4%

#112 · Mini-SWE-Agent

Gemini 2.5 Pro · Gemini 2.5 Pro · Nov 3, 2025

Princeton · Google · API · verified

https://github.com/SWE-agent/mini-swe-agent

26.1%± 1.3%

#113 · Mini-SWE-Agent

Grok Code Fast 1 · Grok Code Fast · Nov 3, 2025

Princeton · xAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

25.8%± 1.3%

#114 · Mini-SWE-Agent

Grok 4 · Grok 4 · Nov 3, 2025

Princeton · xAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

25.4%± 1.5%

#115 · OpenHands

Qwen 3 Coder 480B · Qwen3-Coder 480B A35B · Nov 2, 2025

OpenHands · Alibaba · API · verified

https://github.com/OpenHands/OpenHands

25.4%± 1.3%

#116 · little-coder

Qwen3.6-35B-A3B · Qwen3.6 35B A3B (Reasoning) · May 14, 2026

Itay Inbar · Qwen · API · unverified

https://github.com/itayinbarr/little-coder

24.6%± 1.6%

#117 · Terminus 2

GLM 4.6 · glm-4.6 · Nov 1, 2025

Terminal-Bench · Z.ai · API · verified

https://www.tbench.ai/terminus

24.5%± 1.2%

#118 · Terminus 2

GPT-5-Mini · GPT-5.4 mini · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

24%± 1.3%

#119 · Terminus 2

Qwen 3 Coder 480B · Qwen3-Coder 480B A35B · Nov 1, 2025

Terminal-Bench · Alibaba · API · verified

https://www.tbench.ai/terminus

23.9%± 1.4%

#120 · Terminus 2

Grok 4 · Grok 4 · Oct 31, 2025

Terminal-Bench · xAI · API · verified

https://www.tbench.ai/terminus

23.1%± 1.5%

#121 · little-coder

Qwen3.6-35B-A3B · Qwen3.6 35B A3B (Reasoning) · May 14, 2026

Itay Inbar · Qwen · API · unverified

https://github.com/itayinbarr/little-coder

23%

#122 · Mini-SWE-Agent

GPT-5-Mini · GPT-5.4 mini · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

22.2%± 1.3%

#123 · spoox-o-m

GPT-5-Nano · GPT-5.4 nano · May 15, 2026

TUM · OpenAI · API · unverified

https://github.com/plaume8/spoox

21.8%± 1.4%

#124 · Gemini CLI

Gemini 2.5 Pro · Gemini 2.5 Pro · Nov 4, 2025

Google · Google · API · verified

https://github.com/google-gemini/gemini-cli

19.6%± 1.5%

#125 · Bash Agent

TermiGen-32B · May 14, 2026

UCSB-SURFI · Qwen · API · unverified

https://huggingface.co/UCSB-SURFI/TermiGen-32B

19.3%± 1%

#126 · Terminus 2

GPT-OSS-120B · GPT-OSS 120B · Nov 1, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

18.7%± 1.4%

#127 · Mini-SWE-Agent

Gemini 2.5 Flash · Gemini 2.5 Flash · Nov 3, 2025

Princeton · Google · API · verified

https://github.com/SWE-agent/mini-swe-agent

17.1%± 1.3%

#128 · Terminus 2

AfterQuery-GPT-OSS-20B · Mar 31, 2026

Terminal-Bench · AfterQuery · API · unverified

https://www.tbench.ai/terminus

17%± 1.3%

#129 · Terminus 2

Gemini 2.5 Flash · Gemini 2.5 Flash · Oct 31, 2025

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

16.9%± 1.2%

#130 · OpenHands

Gemini 2.5 Flash · Gemini 2.5 Flash · Nov 2, 2025

OpenHands · Google · API · verified

https://github.com/OpenHands/OpenHands

16.4%± 1.2%

#131 · OpenHands

Gemini 2.5 Pro · Gemini 2.5 Pro · Nov 2, 2025

OpenHands · Google · API · verified

https://github.com/OpenHands/OpenHands

16.4%± 1.4%

#132 · Gemini CLI

Gemini 2.5 Flash · Gemini 2.5 Flash · Nov 4, 2025

Google · Google · API · verified

https://github.com/google-gemini/gemini-cli

15.4%± 1.2%

#133 · Mini-SWE-Agent

GPT-OSS-120B · GPT-OSS 120B · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

14.2%± 1.2%

#134 · Terminus 2

Grok Code Fast 1 · Grok Code Fast · Oct 31, 2025

Terminal-Bench · xAI · API · verified

https://www.tbench.ai/terminus

14.2%± 1.3%

#135 · OpenHands

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 2, 2025

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

13.9%± 1.4%

#136 · Codex CLI

GPT-5-Nano · GPT-5.4 nano · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

11.5%± 1.2%

#137 · OpenHands

GPT-5-Nano · GPT-5.4 nano · Nov 2, 2025

OpenHands · OpenAI · API · verified

https://github.com/OpenHands/OpenHands

9.9%± 1.1%

#138 · little-coder

Qwen3.5-9B · Qwen3.5 9B · May 14, 2026

Itay Inbar · Qwen · API · unverified

https://github.com/itayinbarr/little-coder

9.2%± 1.2%

#139 · Terminus 2

GPT-5-Nano · GPT-5.4 nano · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

7.9%± 1%

#140 · Mini-SWE-Agent

GPT-5-Nano · GPT-5.4 nano · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

7%± 1%

#141 · Mini-SWE-Agent

GPT-OSS-20B · GPT-OSS 20B · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

3.4%± 0.7%

#142 · Terminus 2

GPT-OSS-20B · GPT-OSS 20B · Nov 1, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

3.1%± 0.8%

TerminalBench task registry

tasks

Terminal-Bench 2.0

Official Terminal-Bench 2.0 leaderboard for realistic multi-step terminal tasks.

Source · Terminal-Bench
Version · terminal-bench snapshot 2026-06-24
Scores · 31

Test details

Verified but agingThis is an objective signal, so it is mainly about measurable task performance rather than public taste.

source

Terminal-Bench

metric

Accuracy (%)

judge

Objective

direction

higher better

group id

terminal_bench_2_live

domain

Coding

What it measures vs what it misses

✓ Measures

End-to-end task success on hard terminal workflows that require planning, editing, debugging, and execution.

✗ Misses

IDE-native workflows, code review quality, and non-terminal product engineering work.

Leaderboard · this benchmark version

#1 · GPT-5.5

TERMINAL-BENCH · Apr 23, 2026

Source label: gpt-5.5

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 100%
Last updated: recent
Eligibility: headline eligible
Identity: provider alias (0.92)

82.2%

#2 · GPT-5.3 Codex

TERMINAL-BENCH · Feb 6, 2026

Source label: gpt-5.3-codex

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 96.7%
Last updated: aging
Eligibility: specialized_model
Identity: provider alias (0.92)

75.1%

#3 · Gemini 3 Pro Preview

TERMINAL-BENCH · Jan 6, 2026

verified runtimeexact directBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 93.3%
Last updated: archived
Eligibility: preview_model
Identity: exact (1.00)

69.4%

#4 · Claude Opus 4.6

TERMINAL-BENCH · Feb 6, 2026

verified runtimeexact directBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 90%
Last updated: aging
Eligibility: historical_model
Identity: exact (1.00)

62.9%

#5 · GPT-5.2

TERMINAL-BENCH · Dec 18, 2025

Source label: gpt-5.2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 90%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

62.9%

#6 · GPT-5.1

TERMINAL-BENCH · Nov 24, 2025

Source label: gpt-5.1-codex-max

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 83.3%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

60.4%

#7 · Claude Opus 4.5

TERMINAL-BENCH · Nov 22, 2025

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

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 80%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.5. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.291 percentage points.

57.8%

#8 · Gemini 3 Flash Preview

TERMINAL-BENCH · Jan 7, 2026

verified runtimeexact directBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 76.7%
Last updated: archived
Eligibility: preview_model
Identity: exact (1.00)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Gemini 3 Flash. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.597 percentage points.

51.7%

#9 · GPT-5

TERMINAL-BENCH · Nov 4, 2025

verified runtimeexact direct

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 73.3%
Last updated: archived
Eligibility: headline eligible
Identity: exact (1.00)

49.6%

#10 · GPT-5.4

TERMINAL-BENCH · Nov 4, 2025

Source label: gpt-5

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 73.3%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

49.6%

#11 · Kimi K2.5 (Reasoning)

TERMINAL-BENCH · Feb 4, 2026

Source label: kimi-k2.5

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 66.7%
Last updated: aging
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Kimi K2.5. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.468 percentage points.

43.2%

#12 · Claude Sonnet 4.5

TERMINAL-BENCH · Oct 31, 2025

Source label: claude-sonnet-4-5-20250929

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 63.3%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Sonnet 4.5. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.423 percentage points.

42.8%

#13 · Claude Opus 4.1

TERMINAL-BENCH · Oct 31, 2025

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

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 60%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.1. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.339 percentage points.

38%

#14 · Claude Opus 4

TERMINAL-BENCH · Oct 31, 2025

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

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 60%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.1. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.339 percentage points. Backfilled from Claude Opus 4.1 via approved benchmark identity mapping map-claude-opus-4-to-4-1.

38%

#15 · Claude Opus 4.7

TERMINAL-BENCH · Oct 31, 2025

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

backfilledproxy backfilledBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 60%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.
Identity: benchmark proxy (0.58)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Claude Opus 4.1. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.339 percentage points. Backfilled from Claude Opus 4.1 via approved benchmark identity mapping map-claude-opus-4-7-to-4-1.

38%

#16 · MiniMax-M2.1

TERMINAL-BENCH · Dec 22, 2025

Source label: minimax-m2.1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 50%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

36.6%

#17 · Kimi K2 Thinking

TERMINAL-BENCH · Nov 11, 2025

Source label: moonshotai/Kimi-K2-Thinking

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 46.7%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Kimi K2 Thinking. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.439 percentage points.

35.7%

#18 · glm-4.7

TERMINAL-BENCH · Jan 28, 2026

Source label: glm-4.7

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 43.3%
Last updated: aging
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: GLM 4.7. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.450 percentage points.

33.4%

#19 · Gemini 2.5 Pro

TERMINAL-BENCH · Oct 31, 2025

Source label: gemini-2.5-pro

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 40%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Gemini 2.5 Pro. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.542 percentage points.

32.6%

#20 · GPT-5.4 mini

TERMINAL-BENCH · Nov 4, 2025

Source label: gpt-5-mini

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 36.7%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.53.0). Display model: GPT-5-Mini. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.530 percentage points.

31.9%

#21 · minimax-m2

TERMINAL-BENCH · Nov 1, 2025

Source label: accounts/fireworks/models/minimax-m2

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 33.3%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: MiniMax M2. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.398 percentage points.

30%

#22 · Claude Haiku 4.5

TERMINAL-BENCH · Nov 3, 2025

Source label: claude-haiku-4-5-20251001

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 30%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

29.8%

#23 · Kimi K2 0905

TERMINAL-BENCH · Nov 1, 2025

Source label: moonshotai/Kimi-K2-Instruct-0905

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 26.7%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: Kimi K2 Instruct. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.287 percentage points.

27.8%

#24 · Grok 4

TERMINAL-BENCH · Nov 2, 2025

Source label: grok-4-0709

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 23.3%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

27.2%

#25 · Grok Code Fast

TERMINAL-BENCH · Nov 3, 2025

Source label: grok-code-fast-1

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 20%
Last updated: archived
Eligibility: specialized_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: Grok Code Fast 1. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 1.339 percentage points.

25.8%

#26 · Qwen3-Coder 480B A35B

TERMINAL-BENCH · Nov 2, 2025

Source label: Qwen/Qwen3-Coder-480B-A35B-Instruct-FP8

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 16.7%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: OpenHands (0.60.0). Display model: Qwen 3 Coder 480B. Integration method: API. Agent URL: https://github.com/OpenHands/OpenHands. Reported stderr: 1.320 percentage points.

25.4%

#27 · glm-4.6

TERMINAL-BENCH · Nov 1, 2025

Source label: accounts/fireworks/models/glm-4p6

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 13.3%
Last updated: archived
Eligibility: benchmark_derived_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: GLM 4.6. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.247 percentage points.

24.5%

#28 · GPT-OSS 120B

TERMINAL-BENCH · Nov 1, 2025

Source label: openai/gpt-oss-120b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 10%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Terminus 2 (2.0.0). Display model: GPT-OSS-120B. Integration method: API. Agent URL: https://www.tbench.ai/terminus. Reported stderr: 1.357 percentage points.

18.7%

#29 · Gemini 2.5 Flash

TERMINAL-BENCH · Nov 3, 2025

Source label: gemini-2.5-flash

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 6.7%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: Gemini 2.5 Flash. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 1.271 percentage points.

17.1%

#30 · GPT-5.4 nano

TERMINAL-BENCH · Nov 4, 2025

Source label: gpt-5-nano

verified runtimeexact alias

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 3.3%
Last updated: archived
Eligibility: headline eligible
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Codex CLI (0.53.0). Display model: GPT-5-Nano. Integration method: API. Agent URL: https://developers.openai.com/codex/cli/. Reported stderr: 1.157 percentage points.

11.5%

#31 · GPT-OSS 20B

TERMINAL-BENCH · Nov 3, 2025

Source label: openai/gpt-oss-20b

verified runtimeexact aliasBackground only

Raw row drilldownsource, percentile, eligibility

Source URL: https://www.tbench.ai/leaderboard/terminal-bench/2.0
Percentile: 0%
Last updated: archived
Eligibility: historical_model
Identity: provider alias (0.92)

Parsed from the public Terminal-Bench 2.0 verified leaderboard. Collapse policy: highest verified score per canonical model. Selected agent: Mini-SWE-Agent (unknown). Display model: GPT-OSS-20B. Integration method: API. Agent URL: https://github.com/SWE-agent/mini-swe-agent. Reported stderr: 0.693 percentage points.

3.4%

TerminalBench agent leaderboard

#1 · NexAU-AHE

GPT-5.5 · GPT-5.5 · May 14, 2026

china-qijizhifeng · OpenAI · API · unverified

https://github.com/china-qijizhifeng/agentic-harness-engineering.git

84.7%± 1.1%

#2 · LemonHarness

Gemini 3.1 Pro Preview · Gemini 3.1 Pro Preview · May 14, 2026

LR AILab of Lenovo CTO Org · Google, OpenAI · API · unverified

https://github.com/Open-Lemon/LemonAgent

84.5%± 1.3%

#3 · Capy

GPT-5.5 · GPT-5.5 · May 14, 2026

Capy · OpenAI · API · unverified

https://capy.ai

83.1%± 1.1%

#4 · Codex CLI

GPT-5.5 · GPT-5.5 · Apr 23, 2026

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli

82.2%± 1.1%

#5 · Polaris

Claude Opus 4.7 · Claude Opus 4.7 · May 14, 2026

PolarisOps · Anthropic, Google, OpenAI · API · unverified

https://polarops.ai

82.2%± 1.4%

#6 · WOZCODE

Claude Opus 4.7 · Claude Opus 4.7 · May 14, 2026

WOZCODE · Anthropic · API · unverified

https://github.com/WithWoz/wozcode-plugin

80.2%± 1.1%

#7 · TongAgents

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · Mar 13, 2026

BIGAI · Google · API · unverified

https://www.bigai.ai/

80.2%± 1.3%

#8 · LemonHarness

Gemini 3.1 Pro Preview · Gemini 3.1 Pro Preview · May 14, 2026

LR AILab of Lenovo CTO Org · Google, OpenAI · API · unverified

https://github.com/Open-Lemon/LemonAgent

79.9%± 1.5%

#9 · SageAgent

GPT-5.3-Codex · GPT-5.3 Codex · Mar 13, 2026

OpenSage · OpenAI · API · unverified

https://github.com/opensage-agent

78.4%± 1.1%

#10 · Droid

GPT-5.3-Codex · GPT-5.3 Codex · Feb 24, 2026

Factory · OpenAI · API · unverified

https://factory.ai/

77.3%± 1.1%

#11 · Meta-Harness

Claude Opus 4.6 · Claude Opus 4.6 · May 14, 2026

Stanford IRIS · Anthropic · API · unverified

https://github.com/stanford-iris-lab/meta-harness-tbench2-artifact

76.4%± 1.2%

#12 · CodeBrain-1.5

GPT-5.3-Codex · GPT-5.3 Codex · Feb 10, 2026

Feeling AI · OpenAI · API · unverified

https://feeling-ai.com

75.8%± 1%

#13 · Codelia

GPT-5.3-Codex · GPT-5.3 Codex · May 14, 2026

kousw · OpenAI · API · unverified

https://github.com/kousw/codelia

75.7%± 1.1%

#14 · Capy

Claude Opus 4.6 · Claude Opus 4.6 · Mar 12, 2026

Capy · Anthropic · API · unverified

https://capy.ai

75.3%± 1.2%

#15 · Simple Codex

GPT-5.3-Codex · GPT-5.3 Codex · Feb 6, 2026

OpenAI · OpenAI · API · verified

https://openai.com/codex

75.1%± 1.2%

#16 · Terminus-KIRA

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · Feb 23, 2026

KRAFTON AI · Google · API · unverified

https://www.krafton.ai/en/

74.8%± 1.3%

#17 · Terminus-KIRA

Claude Opus 4.6 · Claude Opus 4.6 · Feb 22, 2026

KRAFTON AI · Anthropic · API · unverified

https://www.krafton.ai/en/

74.7%± 1.3%

#18 · Mux

GPT-5.3-Codex · GPT-5.3 Codex · Mar 6, 2026

Coder · OpenAI · API · unverified

https://github.com/coder/mux

74.6%± 1.3%

#19 · MAYA-V2

Claude 4.6 Opus · Mar 12, 2026

ADYA · Anthropic · API · unverified

https://adya.ai/maya

72.1%± 1.1%

#20 · TongAgents

Claude Opus 4.6 · Claude Opus 4.6 · Feb 22, 2026

Bigai · Anthropic · API · unverified

https://www.bigai.ai/

71.9%± 1.4%

#21 · spoox-o-m

GPT-5.3-Codex · GPT-5.3 Codex · May 15, 2026

TUM · OpenAI · API · unverified

https://github.com/plaume8/spoox

71.5%± 1.3%

#22 · Junie CLI

Claude Opus 4.6 · Claude Opus 4.6 · Mar 7, 2026

JetBrains · Anthropic, Google, Google, OpenAI · API · unverified

https://junie.jetbrains.com

71%± 1.5%

#23 · Droid

Claude Opus 4.6 · Claude Opus 4.6 · Feb 5, 2026

Factory · Anthropic · API · unverified

https://factory.ai/

69.9%± 1.3%

#24 · Ante

Gemini 3 Pro · Gemini 3 Pro Preview · Jan 6, 2026

Antigma Labs · Google · API · verified

https://antigma.ai

69.4%± 1.1%

#25 · IndusAGI Coding Agent

GPT-5.3-Codex · GPT-5.3 Codex · Mar 18, 2026

Varun Israni (SoloVpx) · OpenAI · API · unverified

https://github.com/varun-israni/indusagi-coding-agent

69.1%± 1.2%

#26 · Crux

Claude Opus 4.6 · Claude Opus 4.6 · Feb 23, 2026

Roam · Anthropic · API · unverified

https://github.com/0xDarkMatter/crux-agent

66.9%

#27 · Deep Agents

GPT-5.2-Codex · GPT-5.2 · Feb 12, 2026

LangChain · OpenAI · API · unverified

https://github.com/langchain-ai/deepagents

66.5%± 1.6%

#28 · Mux

Claude Opus 4.6 · Claude Opus 4.6 · Feb 13, 2026

Coder · Anthropic · API · unverified

https://github.com/coder/mux

66.5%± 1.3%

#29 · clnkr

GPT-5.5 · GPT-5.5 · May 14, 2026

clnkr · OpenAI · API · unverified

https://github.com/clnkr-ai/clnkr

66.1%± 1.3%

#30 · SageAgent

Gemini 3 Pro · Gemini 3 Pro Preview · Feb 23, 2026

OpenSage · Google · API · unverified

https://github.com/opensage-agent

65.2%± 1.1%

#31 · Droid

GPT-5.2 · GPT-5.2 · Dec 24, 2025

Factory · OpenAI · API · unverified

https://factory.ai/

64.9%± 1.4%

#32 · Terminus 2

GPT-5.3-Codex · GPT-5.3 Codex · Feb 5, 2026

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

64.7%± 1.4%

#33 · Junie CLI

Gemini 3 Flash · Gemini 3 Flash Preview · Dec 23, 2025

JetBrains · Google · API · unverified

https://junie.jetbrains.com

64.3%± 1.4%

#34 · Droid

Claude Opus 4.5 · Claude Opus 4.5 · Dec 11, 2025

Factory · Anthropic · API · unverified

https://factory.ai/

63.1%± 1.4%

#35 · Codex CLI

GPT-5.2 · GPT-5.2 · Dec 18, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

62.9%± 1.5%

#36 · Terminus 2

Claude Opus 4.6 · Claude Opus 4.6 · Feb 6, 2026

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

62.9%± 1.4%

#37 · CodeBrain-1.5

Gemini 3 Pro · Gemini 3 Pro Preview · Feb 5, 2026

Feeling AI · Google · API · unverified

https://feeling-ai.com

62.2%± 1.3%

#38 · II-Agent

Gemini 3 Pro · Gemini 3 Pro Preview · Dec 23, 2025

Intelligent Internet · Google · API · unverified

https://agent.ii.inc/

61.8%± 1.4%

#39 · hookele

GPT-5.1-Codex-Mini · GPT-5.1 · May 14, 2026

Dmitry Barakhov · OpenAI · API · unverified

https://github.com/sady4850/hookele_coding_agent

61.6%± 1%

#40 · Gemini CLI

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · May 14, 2026

Google · Google · API · unverified

https://github.com/google-gemini/gemini-cli

61.4%± 2.1%

#41 · Warp

Claude Haiku 4.5 · Claude Haiku 4.5 · Dec 12, 2025

Warp · Anthropic, OpenAI · API · unverified

https://www.warp.dev/

61.2%± 1.5%

#42 · Droid

Gemini 3 Pro · Gemini 3 Pro Preview · Dec 24, 2025

Factory · Google · API · unverified

https://factory.ai/

61.1%± 1.4%

#43 · Mux

GPT-5.2 · GPT-5.2 · Jan 17, 2026

Coder · OpenAI · API · unverified

https://github.com/coder/mux

60.7%

#44 · Codex CLI

GPT-5.1-Codex-Max · GPT-5.1 · Nov 24, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

60.4%± 1.4%

#45 · Gemini CLI

Gemini 3.1 Pro · Gemini 3.1 Pro Preview · May 14, 2026

Google · Google · API · unverified

https://github.com/google-gemini/gemini-cli

59.4%± 2.1%

#46 · Letta Code

Claude Opus 4.5 · Claude Opus 4.5 · Dec 17, 2025

Letta · Anthropic · API · unverified

https://www.letta.com/

59.1%± 1.2%

#47 · Warp

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 20, 2025

Warp · Anthropic, Google, OpenAI · API · unverified

https://www.warp.dev/

59.1%± 1.4%

#48 · Abacus AI Desktop

Claude Haiku 4.5 · Claude Haiku 4.5 · Dec 11, 2025

Abacus.AI · Anthropic, Google · API · unverified

https://deepagent-desktop.abacus.ai/

58.4%± 1.4%

#49 · Mux

Claude Opus 4.5 · Claude Opus 4.5 · Jan 17, 2026

Coder · Anthropic · API · unverified

https://github.com/coder/mux

58.4%

#50 · Claude Code

Claude Opus 4.6 · Claude Opus 4.6 · Feb 7, 2026

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

58%± 1.5%

#51 · Crux

GPT-5.1-Codex · GPT-5.1 · Nov 16, 2025

Roam · OpenAI · API · verified

https://github.com/0xDarkMatter/crux-agent

57.8%± 1.5%

#52 · Terminus 2

Claude Opus 4.5 · Claude Opus 4.5 · Nov 22, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

57.8%± 1.3%

#53 · Grok CLI

Grok 4.20 Reasoning · Grok 4.20 · Apr 2, 2026

Superagent · xAI · API · unverified

https://github.com/superagent-ai/grok-cli

57.3%

#54 · Terminus 2

Gemini 3 Pro · Gemini 3 Pro Preview · Nov 21, 2025

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

56.9%± 1.3%

#55 · Letta Code

Gemini 3 Pro · Gemini 3 Pro Preview · Dec 17, 2025

Letta · Google · API · unverified

https://www.letta.com/

56%± 1.5%

#56 · Goose

Claude Opus 4.5 · Claude Opus 4.5 · Dec 11, 2025

Block · Anthropic · API · unverified

https://block.github.io/goose/

54.3%± 1.3%

#57 · Terminus 2

GPT-5.2 · GPT-5.2 · Dec 12, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

54%± 1.5%

#58 · Letta Code

GPT-5.1-Codex · GPT-5.1 · Dec 17, 2025

Letta · OpenAI · API · unverified

https://www.letta.com/

53.5%± 1.4%

#59 · Simplai Agent

Claude Sonnet 4.6 · Claude Sonnet 4.6 · May 14, 2026

SimplAI · Anthropic · API · unverified

https://simplai.ai

53.4%± 1.4%

#60 · Terminus 2

GLM 5 · glm-5 · Feb 23, 2026

Terminal-Bench · Z-AI · API · unverified

https://www.tbench.ai/terminus

52.4%± 1.3%

#61 · Claude Code

Claude Opus 4.5 · Claude Opus 4.5 · Dec 18, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

52.1%± 1.3%

#62 · OpenHands

Claude Opus 4.5 · Claude Opus 4.5 · Jan 4, 2026

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

51.9%± 1.5%

#63 · Terminus 2

Gemini 3 Flash · Gemini 3 Flash Preview · Jan 7, 2026

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

51.7%± 1.6%

#64 · OpenCode

Claude Opus 4.5 · Claude Opus 4.5 · Jan 12, 2026

Anomaly Innovations · Anthropic · API · unverified

https://opencode.ai

51.7%

#65 · Warp

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 11, 2025

Warp · Anthropic, Anthropic, OpenAI · API · unverified

https://www.warp.dev/

50.1%± 1.4%

#66 · Codex CLI

GPT-5 · GPT-5 · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

49.6%± 1.5%

#67 · Terminus 2

GPT-5.1 · GPT-5.1 · Nov 16, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

47.6%± 1.4%

#68 · Gemini CLI

Gemini 3 Flash · Gemini 3 Flash Preview · Mar 6, 2026

Google · Google · API · unverified

https://github.com/google-gemini/gemini-cli

47.4%± 1.5%

#69 · CAMEL-AI

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Dec 24, 2025

CAMEL-AI · Anthropic · API · unverified

https://www.camel-ai.org/

46.5%± 1.2%

#70 · IndusAGI Coding Agent

MiniMax M2.7 · minimax-m2.7 · May 14, 2026

Varun Israni (SoloVpx) · Minimax · API · unverified

https://github.com/varun-israni/indusagi-coding-agent

45.1%

#71 · Codex CLI

GPT-5-Codex · GPT-5.3 Codex · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

44.3%± 1.4%

#72 · OpenHands

GPT-5 · GPT-5 · Nov 2, 2025

OpenHands · OpenAI · API · verified

https://github.com/OpenHands/OpenHands

43.8%± 1.5%

#73 · Terminus 2

GPT-5-Codex · GPT-5.3 Codex · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

43.4%± 1.5%

#74 · Terminus 2

Kimi K2.5 · Kimi K2.5 (Reasoning) · Feb 4, 2026

Terminal-Bench · Kimi · API · verified

https://www.tbench.ai/terminus

43.2%± 1.5%

#75 · Goose

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Dec 11, 2025

Block · Anthropic · API · unverified

https://block.github.io/goose/

43.1%± 1.3%

#76 · Crux

GPT-5.1-Codex-Mini · GPT-5.1 · Nov 17, 2025

Roam · OpenAI · API · verified

https://github.com/0xDarkMatter/crux-agent

43.1%± 1.5%

#77 · Harness Agent

MiniMax M2.7 Highspeed · May 14, 2026

lazyFrogLOL · MiniMax · API · unverified

https://github.com/lazyFrogLOL/Harness_Engineering

42.9%± 1.5%

#78 · Terminus 2

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Oct 31, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

42.8%± 1.4%

#79 · MAYA-V2

Claude 4.5 Sonnet · Claude Sonnet 4.5 · Jan 4, 2026

ADYA · Anthropic · API · unverified

https://adya.ai/maya

42.7%

#80 · cchuter

minimax-m2.5 · minimax-m2.5 · Mar 30, 2026

teamblobfish.com · minimax · API · unverified

https://github.com/cchuter/blobfish

42.7%± 1.4%

#81 · OpenHands

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Nov 2, 2025

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

42.6%± 1.4%

#82 · Mini-SWE-Agent

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Nov 3, 2025

Princeton · Anthropic · API · verified

https://github.com/SWE-agent/mini-swe-agent

42.5%± 1.4%

#83 · Terminus 2

Minimax m2.5 · minimax-m2.5 · Feb 23, 2026

Terminal-Bench · Minimax · API · unverified

https://www.tbench.ai/terminus

42.2%± 1.3%

#84 · Mini-SWE-Agent

GPT-5-Codex · GPT-5.3 Codex · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

41.3%± 1.4%

#85 · Claude Code

Claude Sonnet 4.5 · Claude Sonnet 4.5 · Nov 4, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

40.1%± 1.5%

#86 · Terminus 2

DeepSeek-V3.2 · DeepSeek Chat · Feb 10, 2026

Terminal-Bench · DeepSeek · API · unverified

https://www.tbench.ai/terminus

39.6%± 1.4%

#87 · Terminus 2

Claude Opus 4.1 · Claude Opus 4.1 · Oct 31, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

38%± 1.3%

#88 · OpenHands

Claude Opus 4.1 · Claude Opus 4.1 · Nov 2, 2025

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

36.9%± 1.4%

#89 · Terminus 2

GPT-5.1-Codex · GPT-5.1 · Nov 17, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

36.9%± 1.7%

#90 · Crux

MiniMax M2.1 · MiniMax-M2.1 · Dec 22, 2025

Roam · MiniMax · API · verified

https://github.com/0xDarkMatter/crux-agent

36.6%± 1.5%

#91 · Terminus 2

Kimi K2 Thinking · Kimi K2 Thinking · Nov 11, 2025

Terminal-Bench · Moonshot AI · API · verified

https://www.tbench.ai/terminus

35.7%± 1.4%

#92 · Goose

Claude Haiku 4.5 · Claude Haiku 4.5 · Dec 11, 2025

Block · Anthropic · API · unverified

https://block.github.io/goose/

35.5%± 1.5%

#93 · Terminus 2

GPT-5 · GPT-5 · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

35.2%± 1.6%

#94 · Mini-SWE-Agent

Claude Opus 4.1 · Claude Opus 4.1 · Nov 3, 2025

Princeton · Anthropic · API · verified

https://github.com/SWE-agent/mini-swe-agent

35.1%± 1.3%

#95 · Claude Code

Claude Opus 4.1 · Claude Opus 4.1 · Nov 4, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

34.8%± 1.5%

#96 · spoox-o-m

GPT-5-Mini · GPT-5.4 mini · Dec 24, 2025

TUM · OpenAI · API · unverified

https://github.com/plaume8/spoox

34.8%± 1.4%

#97 · Mini-SWE-Agent

GPT-5 · GPT-5 · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

33.9%± 1.5%

#98 · Terminus 2

GLM 4.7 · glm-4.7 · Jan 28, 2026

Terminal-Bench · Z-AI · API · verified

https://www.tbench.ai/terminus

33.4%± 1.4%

#99 · Crux

GLM 4.7 · glm-4.7 · Feb 8, 2026

Roam · Z-AI · API · unverified

https://github.com/0xDarkMatter/crux-agent

33.3%± 1.3%

#100 · Terminus 2

Gemini 2.5 Pro · Gemini 2.5 Pro · Oct 31, 2025

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

32.6%± 1.5%

#101 · Codex CLI

GPT-5-Mini · GPT-5.4 mini · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

31.9%± 1.5%

#102 · Terminus 2

MiniMax M2 · minimax-m2 · Nov 1, 2025

Terminal-Bench · MiniMax · API · verified

https://www.tbench.ai/terminus

30%± 1.4%

#103 · Mini-SWE-Agent

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 3, 2025

Princeton · Anthropic · API · verified

https://github.com/SWE-agent/mini-swe-agent

29.8%± 1.3%

#104 · Terminus 2

MiniMax M2.1 · MiniMax-M2.1 · Dec 23, 2025

Terminal-Bench · MiniMax · API · verified

https://www.tbench.ai/terminus

29.2%± 1.5%

#105 · OpenHands

GPT-5-Mini · GPT-5.4 mini · Nov 2, 2025

OpenHands · OpenAI · API · verified

https://github.com/OpenHands/OpenHands

29.2%± 1.4%

#106 · Terminus 2

Claude Haiku 4.5 · Claude Haiku 4.5 · Oct 31, 2025

Terminal-Bench · Anthropic · API · verified

https://www.tbench.ai/terminus

28.3%± 1.5%

#107 · Terminus 2

Kimi K2 Instruct · Kimi K2 0905 · Nov 1, 2025

Terminal-Bench · Moonshot AI · API · verified

https://www.tbench.ai/terminus

27.8%± 1.3%

#108 · Claude Code

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 4, 2025

Anthropic · Anthropic · API · verified

https://www.claude.com/product/claude-code

27.5%± 1.4%

#109 · OpenHands

Grok 4 · Grok 4 · Nov 2, 2025

OpenHands · xAI · API · verified

https://github.com/OpenHands/OpenHands

27.2%± 1.6%

#110 · Dakou Agent

Qwen 3 Coder 480B · Qwen3-Coder 480B A35B · Dec 28, 2025

iflow · Alibaba · API · unverified

https://dakou.iflow.cn/

27.2%± 1.3%

#111 · OpenHands

Kimi K2 Instruct · Kimi K2 0905 · Nov 2, 2025

OpenHands · Moonshot AI · API · verified

https://github.com/OpenHands/OpenHands

26.7%± 1.4%

#112 · Mini-SWE-Agent

Gemini 2.5 Pro · Gemini 2.5 Pro · Nov 3, 2025

Princeton · Google · API · verified

https://github.com/SWE-agent/mini-swe-agent

26.1%± 1.3%

#113 · Mini-SWE-Agent

Grok Code Fast 1 · Grok Code Fast · Nov 3, 2025

Princeton · xAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

25.8%± 1.3%

#114 · Mini-SWE-Agent

Grok 4 · Grok 4 · Nov 3, 2025

Princeton · xAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

25.4%± 1.5%

#115 · OpenHands

Qwen 3 Coder 480B · Qwen3-Coder 480B A35B · Nov 2, 2025

OpenHands · Alibaba · API · verified

https://github.com/OpenHands/OpenHands

25.4%± 1.3%

#116 · little-coder

Qwen3.6-35B-A3B · Qwen3.6 35B A3B (Reasoning) · May 14, 2026

Itay Inbar · Qwen · API · unverified

https://github.com/itayinbarr/little-coder

24.6%± 1.6%

#117 · Terminus 2

GLM 4.6 · glm-4.6 · Nov 1, 2025

Terminal-Bench · Z.ai · API · verified

https://www.tbench.ai/terminus

24.5%± 1.2%

#118 · Terminus 2

GPT-5-Mini · GPT-5.4 mini · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

24%± 1.3%

#119 · Terminus 2

Qwen 3 Coder 480B · Qwen3-Coder 480B A35B · Nov 1, 2025

Terminal-Bench · Alibaba · API · verified

https://www.tbench.ai/terminus

23.9%± 1.4%

#120 · Terminus 2

Grok 4 · Grok 4 · Oct 31, 2025

Terminal-Bench · xAI · API · verified

https://www.tbench.ai/terminus

23.1%± 1.5%

#121 · little-coder

Qwen3.6-35B-A3B · Qwen3.6 35B A3B (Reasoning) · May 14, 2026

Itay Inbar · Qwen · API · unverified

https://github.com/itayinbarr/little-coder

23%

#122 · Mini-SWE-Agent

GPT-5-Mini · GPT-5.4 mini · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

22.2%± 1.3%

#123 · spoox-o-m

GPT-5-Nano · GPT-5.4 nano · May 15, 2026

TUM · OpenAI · API · unverified

https://github.com/plaume8/spoox

21.8%± 1.4%

#124 · Gemini CLI

Gemini 2.5 Pro · Gemini 2.5 Pro · Nov 4, 2025

Google · Google · API · verified

https://github.com/google-gemini/gemini-cli

19.6%± 1.5%

#125 · Bash Agent

TermiGen-32B · May 14, 2026

UCSB-SURFI · Qwen · API · unverified

https://huggingface.co/UCSB-SURFI/TermiGen-32B

19.3%± 1%

#126 · Terminus 2

GPT-OSS-120B · GPT-OSS 120B · Nov 1, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

18.7%± 1.4%

#127 · Mini-SWE-Agent

Gemini 2.5 Flash · Gemini 2.5 Flash · Nov 3, 2025

Princeton · Google · API · verified

https://github.com/SWE-agent/mini-swe-agent

17.1%± 1.3%

#128 · Terminus 2

AfterQuery-GPT-OSS-20B · Mar 31, 2026

Terminal-Bench · AfterQuery · API · unverified

https://www.tbench.ai/terminus

17%± 1.3%

#129 · Terminus 2

Gemini 2.5 Flash · Gemini 2.5 Flash · Oct 31, 2025

Terminal-Bench · Google · API · verified

https://www.tbench.ai/terminus

16.9%± 1.2%

#130 · OpenHands

Gemini 2.5 Flash · Gemini 2.5 Flash · Nov 2, 2025

OpenHands · Google · API · verified

https://github.com/OpenHands/OpenHands

16.4%± 1.2%

#131 · OpenHands

Gemini 2.5 Pro · Gemini 2.5 Pro · Nov 2, 2025

OpenHands · Google · API · verified

https://github.com/OpenHands/OpenHands

16.4%± 1.4%

#132 · Gemini CLI

Gemini 2.5 Flash · Gemini 2.5 Flash · Nov 4, 2025

Google · Google · API · verified

https://github.com/google-gemini/gemini-cli

15.4%± 1.2%

#133 · Mini-SWE-Agent

GPT-OSS-120B · GPT-OSS 120B · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

14.2%± 1.2%

#134 · Terminus 2

Grok Code Fast 1 · Grok Code Fast · Oct 31, 2025

Terminal-Bench · xAI · API · verified

https://www.tbench.ai/terminus

14.2%± 1.3%

#135 · OpenHands

Claude Haiku 4.5 · Claude Haiku 4.5 · Nov 2, 2025

OpenHands · Anthropic · API · verified

https://github.com/OpenHands/OpenHands

13.9%± 1.4%

#136 · Codex CLI

GPT-5-Nano · GPT-5.4 nano · Nov 4, 2025

OpenAI · OpenAI · API · verified

https://developers.openai.com/codex/cli/

11.5%± 1.2%

#137 · OpenHands

GPT-5-Nano · GPT-5.4 nano · Nov 2, 2025

OpenHands · OpenAI · API · verified

https://github.com/OpenHands/OpenHands

9.9%± 1.1%

#138 · little-coder

Qwen3.5-9B · Qwen3.5 9B · May 14, 2026

Itay Inbar · Qwen · API · unverified

https://github.com/itayinbarr/little-coder

9.2%± 1.2%

#139 · Terminus 2

GPT-5-Nano · GPT-5.4 nano · Oct 31, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

7.9%± 1%

#140 · Mini-SWE-Agent

GPT-5-Nano · GPT-5.4 nano · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

7%± 1%

#141 · Mini-SWE-Agent

GPT-OSS-20B · GPT-OSS 20B · Nov 3, 2025

Princeton · OpenAI · API · verified

https://github.com/SWE-agent/mini-swe-agent

3.4%± 0.7%

#142 · Terminus 2

GPT-OSS-20B · GPT-OSS 20B · Nov 1, 2025

Terminal-Bench · OpenAI · API · verified

https://www.tbench.ai/terminus

3.1%± 0.8%

TerminalBench task registry

tasks

Terminal-Bench 2.0

Test details

What it measures vs what it misses

✓ Measures

✗ Misses

Leaderboard · this benchmark version

TerminalBench agent leaderboard

TerminalBench task registry

Loading benchmark evidence.

Terminal-Bench 2.0

Test details

What it measures vs what it misses

✓ Measures

✗ Misses

Leaderboard · this benchmark version

TerminalBench agent leaderboard

TerminalBench task registry