Agentic · terminal
Terminal-Bench 4.1 leaderboard
As of 2026-11-18, the highest measured score on Terminal-Bench 4.1 is 65.8% by Claude Fable 5.2. 103 more models have estimated scores, calibrated from the benchmarks they were measured on.
Measured scores: benchlm.ai.
| # | Model | Score | Source |
|---|---|---|---|
| 1 | Claude Fable 5.2 | 65.8% | measured |
| 2 | Claude Opus 5.5 | 65.8% | measured |
| 3 | GPT-6.2 Astra | 65.0% | estimated ± 5.5 pp, low confidence |
| 4 | Gemini 4 Carbon | 64.6% | measured |
| 5 | Claude Mythos 5.2 | 64.6% | estimated ± 5.0 pp, medium confidence |
| 6 | Claude Sonnet 5.5 | 63.9% | measured |
| 7 | GPT-6.2 Sol | 63.2% | estimated ± 5.5 pp, medium confidence |
| 8 | GPT-6.1 Sol | 60.9% | measured |
| 9 | Claude Mythos 5.1 | 60.9% | estimated ± 5.0 pp, medium confidence |
| 10 | GPT-6 Astra | 60.3% | measured |
| 11 | Kimi K3.1 Pro | 59.7% | measured |
| 12 | Claude Fable 5.1 | 57.6% | measured |
| 13 | Muse Wonder Full | 56.5% | estimated ± 6.4 pp, medium confidence |
| 14 | Gemini 4 Hydrogen | 56.4% | measured |
| 15 | Gemini 4 Argon | 55.5% | measured |
| 16 | Claude Opus 5 | 53.3% | measured |
| 17 | GPT-6.1 Terra | 53.3% | measured |
| 18 | Pareto 26.9 | 51.7% | estimated ± 5.0 pp, medium confidence |
| 19 | Pareto 26.10 Preview | 51.4% | estimated ± 5.0 pp, medium confidence |
| 20 | GPT-6 Sol | 50.3% | measured |
| 21 | Qwen4 Max | 48.3% | estimated ± 5.0 pp, medium confidence |
| 22 | Muse Wonder | 47.0% | measured |
| 23 | ERNIE 5.5 Flash Preview | 45.4% | estimated ± 5.0 pp, medium confidence |
| 24 | Claude Fable 5 | 44.9% | measured |
| 25 | Ling 3.1 Flash | 44.7% | estimated ± 5.0 pp, medium confidence |
| 26 | GPT-5.5 | 41.4% | estimated ± 9.2 pp, low confidence |
| 27 | GLM-5.3 | 41.2% | measured |
| 28 | Claude Haiku 5.5 | 40.0% | measured |
| 29 | Grok 4.7 | 38.8% | measured |
| 30 | DeepSeek V4.1 Flash 1017 | 38.5% | measured |
| 31 | GPT-5.6 Sol | 38.5% | measured |
| 32 | Hy4 | 37.9% | measured |
| 33 | GLM-5.3-Flash | 37.0% | measured |
| 34 | LongCat-2.5 | 32.9% | estimated ± 5.0 pp, medium confidence |
| 35 | MiniMax M3.1 Flash | 32.3% | estimated ± 5.0 pp, medium confidence |
| 36 | Muse Ember | 30.6% | measured |
| 37 | Mistral Large 4 | 27.6% | measured |
| 38 | MiMo-V2.6-Pro | 27.0% | measured |
| 39 | Step 5 Preview | 26.0% | estimated ± 5.0 pp, medium confidence |
| 40 | Claude Mythos 5 | 24.9% | estimated ± 10.9 pp, low confidence |
| 41 | Claude Opus 4.8 | 23.6% | measured |
| 42 | Qwen3.8 Max | 23.5% | estimated ± 10.9 pp, low confidence |
| 43 | Kimi K3 | 23.0% | measured |
| 44 | Gemini 3.6 Flash | 22.8% | estimated ± 14.9 pp, low confidence |
| 45 | Hy4 preview | 22.3% | estimated ± 10.9 pp, low confidence |
| 46 | Ember-1 | 21.8% | measured |
| 47 | MAI Code 2 | 21.6% | estimated ± 10.9 pp, low confidence |
| 48 | GPT-5.6 Terra | 21.5% | measured |
| 49 | SWE-2 | 20.3% | estimated ± 5.0 pp, medium confidence |
| 50 | Ultramarine 1 Flash | 20.2% | estimated ± 10.9 pp, low confidence |
| 51 | Muse Spark 1.2 | 19.9% | estimated ± 10.9 pp, low confidence |
| 52 | DeepSeek V4 Flash 0731 | 19.7% | estimated ± 10.9 pp, low confidence |
| 53 | Grok 4.6 | 19.7% | measured |
| 54 | Sakana Fugu-Ultra | 19.1% | estimated ± 10.9 pp, low confidence |
| 55 | SWE-1.7 | 18.6% | estimated ± 10.9 pp, low confidence |
| 56 | Gemini 3.8 Flash | 18.2% | measured |
| 57 | GPT-5.6 Luna | 17.6% | measured |
| 58 | Ripple-65B Preview | 17.6% | estimated ± 10.9 pp, low confidence |
| 59 | Claude Opus 4.7 | 17.4% | estimated ± 14.9 pp, low confidence |
| 60 | Sakana Fugu | 17.3% | estimated ± 10.9 pp, low confidence |
| 61 | Muse Spark 1.1 | 17.1% | estimated ± 10.9 pp, low confidence |
| 62 | Particle-1 | 17.0% | estimated ± 10.9 pp, low confidence |
| 63 | Inkling 2 Small | 16.4% | estimated ± 6.4 pp, medium confidence |
| 64 | SWE-2 Flash | 16.3% | estimated ± 5.0 pp, medium confidence |
| 65 | MiMo-V2.6-Flash | 16.1% | measured |
| 66 | Beam 1.0 501B-A23B FP8 | 16.0% | estimated ± 10.9 pp, low confidence |
| 67 | Hy4 flash | 16.0% | estimated ± 5.0 pp, medium confidence |
| 68 | Beam 1.0 501B-A23B NVFP4 | 15.6% | estimated ± 10.9 pp, low confidence |
| 69 | Atria Dawn Preview | 15.5% | estimated ± 10.9 pp, low confidence |
| 70 | MAI Code 2 Flash | 15.5% | estimated ± 10.9 pp, low confidence |
| 71 | GPT-6 Luna | 15.2% | measured |
| 72 | Muse Spark 1.3 | 14.9% | measured |
| 73 | Ornith-1.0-397B | 14.7% | estimated ± 10.9 pp, low confidence |
| 74 | Gemini 3.5 Flash | 13.4% | estimated ± 10.9 pp, low confidence |
| 75 | Gemini 3.7 Flash | 12.4% | measured |
| 76 | Grok 4.5 | 12.4% | measured |
| 77 | dots3-note Preview | 12.4% | estimated ± 10.9 pp, low confidence |
| 78 | MiMo-V2.5 | 12.2% | estimated ± 14.9 pp, low confidence |
| 79 | Claude Sonnet 5 | 12.1% | measured |
| 80 | GPT-5.4 | 11.1% | estimated ± 9.2 pp, low confidence |
| 81 | Jais-3-375B-A23B | 11.1% | estimated ± 10.9 pp, low confidence |
| 82 | Composer 2.5 | 11.1% | estimated ± 14.9 pp, low confidence |
| 83 | Ornith-1.5-397B | 11.1% | estimated ± 4.8 pp, medium confidence |
| 84 | Claude Sonnet 4.6 | 10.5% | estimated ± 14.9 pp, low confidence |
| 85 | GLM-5.1 | 10.4% | estimated ± 14.9 pp, low confidence |
| 86 | OrcaCyber-Nano-1.0 | 9.8% | estimated ± 10.9 pp, low confidence |
| 87 | Qwen3.8-27B | 9.6% | estimated ± 6.4 pp, medium confidence |
| 88 | Gemini 3 Flash | 9.2% | estimated ± 14.9 pp, low confidence |
| 89 | Seed 2.1 Pro | 8.4% | estimated ± 10.9 pp, low confidence |
| 90 | Apodex 1.1 | 8.2% | estimated ± 10.9 pp, low confidence |
| 91 | Laguna S 2.1 | 7.6% | estimated ± 10.9 pp, low confidence |
| 92 | MiniMax M3 | 7.6% | measured |
| 93 | MiniMax M2.7 | 7.4% | estimated ± 14.9 pp, low confidence |
| 94 | Quasar 438B | 6.8% | estimated ± 10.9 pp, low confidence |
| 95 | DeepSeek V4 Pro | 6.7% | measured |
| 96 | Beam 1.0 501B-A23B | 6.4% | measured |
| 97 | Grok 4.20 | 6.2% | estimated ± 14.9 pp, low confidence |
| 98 | Claude Haiku 4.5 | 6.1% | estimated ± 14.9 pp, low confidence |
| 99 | Grok 4.3 | 5.6% | estimated ± 14.9 pp, low confidence |
| 100 | GPT-5.4 nano | 5.5% | estimated ± 14.9 pp, low confidence |
| 101 | Seed 2.1 Turbo | 5.1% | estimated ± 10.9 pp, low confidence |
| 102 | Mistral Medium 3.5 128B | 5.0% | estimated ± 14.9 pp, low confidence |
| 103 | Jais-3-MoVA-36B-A4B | 4.8% | estimated ± 10.9 pp, low confidence |
| 104 | Mercury 2.5 | 4.1% | estimated ± 14.9 pp, low confidence |
| 105 | Gemini 3.1 Flash-Lite | 4.0% | estimated ± 14.9 pp, low confidence |
| 106 | Laguna M.1 | 4.0% | estimated ± 14.9 pp, low confidence |
| 107 | Gemini 3.1 Pro | 2.7% | measured |
| 108 | Pokee-Isaac 28B | 2.7% | estimated ± 10.9 pp, low confidence |
| 109 | Laguna XS.2 | 2.7% | estimated ± 14.9 pp, low confidence |
| 110 | Qwen3.7 Max | 2.2% | estimated ± 9.2 pp, low confidence |
| 111 | Claude Opus 4.6 | 2.2% | estimated ± 9.2 pp, low confidence |
| 112 | Claude Opus 4.5 Thinking | 2.2% | estimated ± 9.2 pp, low confidence |
| 113 | GPT-5.2 | 2.2% | estimated ± 9.2 pp, low confidence |
| 114 | Qwen3.7 Plus | 2.2% | estimated ± 9.2 pp, low confidence |
| 115 | GPT-5.1 | 2.2% | estimated ± 9.2 pp, low confidence |
| 116 | Muse Spark | 2.2% | estimated ± 9.2 pp, low confidence |
| 117 | Kimi K2.7 Code | 2.2% | estimated ± 9.2 pp, low confidence |
| 118 | Kimi K2.6 | 2.2% | estimated ± 9.2 pp, low confidence |
| 119 | Qwen 3.6 Max (preview) | 2.2% | estimated ± 9.2 pp, low confidence |
| 120 | Qwen3.6 Plus | 2.2% | estimated ± 9.2 pp, low confidence |
| 121 | GLM-5 | 2.2% | estimated ± 9.2 pp, low confidence |
| 122 | MiMo-V2.5-Pro | 2.2% | estimated ± 9.2 pp, low confidence |
| 123 | Ornith-1.0-35B | 1.8% | estimated ± 10.9 pp, low confidence |
| 124 | GPT-5.4 mini | 1.8% | measured |
| 125 | Command A+ | 1.6% | estimated ± 14.9 pp, low confidence |
| 126 | Jais-3-32B | 1.3% | estimated ± 10.9 pp, low confidence |
| 127 | Inkling-Small | 0.9% | measured |
| 128 | MAI-Code-1.1-Flash | 0.6% | estimated ± 10.9 pp, low confidence |
| 129 | Nemotron 3.5 Super VL 120B A12B | 0.3% | measured |
| 130 | A.X K2 | 0.0% | estimated ± 10.9 pp, low confidence |
| 131 | Flint | 0.0% | estimated ± 10.9 pp, low confidence |
| 132 | Gemini 3.5 Flash-Lite | 0.0% | estimated ± 10.9 pp, low confidence |
| 133 | GLM-5.2 | 0.0% | estimated ± 4.8 pp, medium confidence |
| 134 | Inkling | 0.0% | measured |
| 135 | Jais-3-3.7B | 0.0% | estimated ± 10.9 pp, low confidence |
| 136 | Jais-3-7B | 0.0% | estimated ± 10.9 pp, low confidence |
| 137 | K-EXAONE 2.0 | 0.0% | estimated ± 10.9 pp, low confidence |
| 138 | Ling 3.0 Flash | 0.0% | estimated ± 10.9 pp, low confidence |
| 139 | Muse Glimmer 30B | 0.0% | estimated ± 10.9 pp, low confidence |
| 140 | Nemotron 3.5 Bolt 74B-A9B | 0.0% | estimated ± 10.9 pp, low confidence |
| 141 | Nemotron 3.5 Bolt 74B-A9B BF16 | 0.0% | estimated ± 10.9 pp, low confidence |
| 142 | Nemotron 3.5 Super VL 120B A12B NVFP4 | 0.0% | estimated ± 10.9 pp, low confidence |
| 143 | Nemotron 3 Ultra | 0.0% | estimated ± 10.9 pp, low confidence |
| 144 | Ornith-1.0-9B | 0.0% | estimated ± 10.9 pp, low confidence |
| 145 | Ornith-1.5-35B-A3B | 0.0% | estimated ± 4.8 pp, medium confidence |
| 146 | Ornith-1.5-9B | 0.0% | estimated ± 10.9 pp, low confidence |
| 147 | Solar Pro 4 | 0.0% | estimated ± 10.9 pp, low confidence |
| 148 | Step 3.7 Flash | 0.0% | estimated ± 10.9 pp, low confidence |
| 149 | Ternary Bonsai 2 27B | 0.0% | estimated ± 10.9 pp, low confidence |
All leaderboards
Agentic · terminal
- AA Terminal-Bench 2.1
- AA Terminal-Bench 4.0
- Terminal-Bench 4.0
- Terminal-Bench 2.0
- Terminal-Bench 2.1
- Terminal-Bench 3.0
- Terminal-Bench Hard
- Terminal-Bench-Science 0.1
- Terminal-Bench 4.1
- Terminal-Bench 2.1 (Vals AI)
Agentic · tools
- AA Agentic Index
- AA-AnalystAgent
- AA AutomationBench
- CWE-Bench-AA
- AA Cyber Index
- DeepsecBench-AA
- AA EnterpriseOps-Gym
- GDP.pdf
- AA Harvey LAB v1.0
- Harvey LAB-AA v1.1
- AA ITBench
- AA Tau3 Banking
- Agents' Last Exam
- APEX-Agents
- APEX-Agents-AA
- ApprenticeBench
- AutomationBench
- BFCL v4
- BrowseComp
- Claw-Eval
- CWE-bench v1
- CWE-bench
- CyberGym
- DeepPlanning
- DeepSearchQA
- DRACO
- ExploitGym
- Finance Agent v2
- GDPval-AA
- Gert Labs
- Hermes Bench
- Hermes Index
- HLE w/ tools
- JobBench
- LifeVal
- MazeBench
- MCP Atlas
- MCP-Tasks
- MKA
- OSWorld 2.1
- OSWorld 2.1 Offline
- OSWorld-Verified
- OSWorld 2.0
- PinchBench
- QwenClawBench
- ResearchClawBench
- SkillsBench
- τ²-bench results
- τ³-bench results
- Toolathlon
- Toolathlon-Verified
- VITA-Bench
- WideResearch
Coding
- AA Coding Index
- AA LiveCodeBench
- AA-SciCode
- Bug Hunt Bench
- CursorBench 3.1
- CursorBench 3.2
- CursorBench 4.0
- DeepSWE
- FrontierCode 1.1 Main
- FrontierCode 2.0
- FrontierCode 1.1 Extended
- FrontierSWE v2
- LiveCodeBench
- LiveCodeBench Pro
- LiveCodeBench v6
- NL2Repo
- OpenHarmony Bench v1.2
- OpenHarmony Bench v1.0
- PostTrainBench
- PostTrainBench v1
- PostTrainBench v1.1
- ProgramBench
- React Native Evals
- ScarfBench
- SciCode
- SWE-Atlas Codebase QnA
- SWE-bench Pro
- SWE-bench Verified
- SWE-Rebench
- SWE-Marathon
- Vals LiveCodeBench
- Vals SWE-bench
- Vibe Code Bench
- VulcanBench Frontier v4
- VulcanBench v3
Math
- AIME 2024
- AIME 2025
- AIME26
- FrontierMath v2 (Tier 4)
- FrontierMath v2 (Tiers 1-3)
- HMMT Feb 2025
- HMMT Feb 2026
- HMMT Nov 2025
- IMOAnswerBench
- MATH-500
- MMAnswerBench
Knowledge & reasoning
- AA-GPQA Diamond
- HealthBench Professional (length-adjusted, Artificial Analysis)
- HealthBench Professional (raw, Artificial Analysis)
- AA-HLE
- AA MMLU-Pro
- ARC-AGI-2
- ARC-AGI-3 Provider Adapter
- ARC-AGI-1
- ARC-AGI-3
- Artificial Analysis Intelligence Index
- BBH
- BioBasicsBench
- BioMysteryBench (human-difficult)
- BioMysteryBench (human-solvable)
- C-Eval
- CritPt
- GPQA
- GPQA Diamond
- HealthBench (raw)
- HealthBench Hard
- HealthBench (length-adjusted)
- HealthBench Professional
- HealthBench Professional (raw)
- HLE
- HLE w/o tools
- HLE-Verified
- JevBench 1.4
- LABBench2
- MedXpertQA (Text)
- MMLU
- MMLU-Pro
- MMLU-Pro (Arcee)
- MMLU-Redux
- MMMLU
- AA-Omniscience Accuracy
- SimpleQA
- STEMBench
- SuperGPQA
- Vals GPQA Diamond
- Vals MMLU-Pro
Instruction following
Multilingual
Vision & documents
- AA-MMMU-Pro
- AI2D_TEST
- BabyVision
- BabyVision w/ Python
- BenchCAD Vision2Code (no tools)
- BenchCAD Vision2Code (tools)
- CC-OCR
- Chartography (no tools)
- Chartography (tools)
- CharXiv
- CharXiv w/o tools
- CountBench
- DynaMath
- ERQA
- LVBench
- MathVision
- MathVision w/ Python
- MedXpertQA (MM)
- MLVU (M-Avg)
- MMMU
- MMMU-Pro
- MMMU-Pro w/ Python
- MMVU
- OCRBench V2
- OfficeQA
- OfficeQA Pro
- OmniDocBench 1.5
- ParseBench-1
- ParseBench-2
- PerceptionBench
- RealWorldQA
- RefCOCO (avg)
- ScreenSpot Pro
- SimpleVQA
- SWE-bench Multimodal
- Video-MME (w/o subtitle)
- Video-MME (with subtitle)
- VideoMMMU
- V*
- We-Math
- ZeroBench
Long context
General
- HLE-MultiChoice (Decision Index)
- Artificial Analysis Decision Index v1.0
- Artificial Analysis Decision Index
- When2Call (Decision Index)
- WildGuardTest (Decision Index)
- BabaIsBench
- BinaryAudit
- BlenderWorld (text-only)
- Chess Puzzles
- CompileBench
- Cookie-Bench
- DAYJOB
- DAYJOB Healthcare
- Decision Index Full Score
- EBR-bench
- Epoch Automation Reports
- ExploitBench
- ExploitBench Internal Port
- FACTS Multimodal
- FailureSensorIQ
- FrontierPhysics
- Furniture Assembly
- Decision Index Arts
- Decision Index Knowledge
- Decision Index Language
- Decision Index Retrieval
- Decision Index Tools
- MaxxBench
- MirrorCode
- Mystery Game Puzzles
- NYT Connections
- OTelBench
- SEC-Bench Pro
- SkillsBench official leaderboard (BenchFlow)
- SpireClimb
- StealthMark