# LLM benchmark scores, with the gaps filled

LLM benchmark leaderboard · gaps filled · https://benchgap.net/ · data retrieved 2026-10-29

benchgap is an LLM benchmark leaderboard that fills in the missing scores. Most models are only ever run on a handful of benchmarks, so benchgap calibrates benchmarks against each other on the models measured on both, then estimates each missing score with its cross-validated error and a confidence level. Measured and estimated scores are always marked apart.

As of 2026-10-29: 589 models, 195 benchmark versions, 6892 measured and 21935 estimated scores (3781 high, 7914 medium and 10240 low confidence). Measured scores come from public leaderboards (artificial-analysis, published, vals-ai).

## Terminal-Bench 4.0 leaderboard

As of 2026-10-29, the highest measured score on Terminal-Bench 4.0 is 70.6% by Claude Sonnet 5.5. 11 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: [benchlm.ai](https://benchlm.ai/benchmarks/terminal-bench-4).

| # | Model | Score | Source |
|---|---|---|---|
| 1 | [Claude Sonnet 5.5](https://benchgap.net/model/claude-sonnet-5-5.md) | 70.6% | measured |
| 2 | [Claude Opus 5.5](https://benchgap.net/model/claude-opus-5-5.md) | 66.4% | measured |
| 3 | [Kimi K3.1 Pro](https://benchgap.net/model/kimi-3-1-pro.md) | 65.7% | measured |
| 4 | [Gemini 4 Carbon](https://benchgap.net/model/gemini-4-carbon.md) | 64.6% | estimated ± 5.7 pp, medium confidence |
| 5 | [Claude Mythos 5.1](https://benchgap.net/model/claude-mythos-5-1.md) | 60.9% | measured |
| 6 | [Muse Wonder Full](https://benchgap.net/model/muse-wonder-full.md) | 58.9% | estimated ± 5.7 pp, medium confidence |
| 7 | [GPT-6.1 Sol](https://benchgap.net/model/gpt-6-1-sol.md) | 58.2% | estimated ± 5.7 pp, medium confidence |
| 8 | [GPT-6 Astra](https://benchgap.net/model/gpt-6-astra.md) | 57.9% | measured |
| 9 | [Gemini 4 Argon](https://benchgap.net/model/gemini-4-argon.md) | 57.4% | measured |
| 10 | [Claude Fable 5.1](https://benchgap.net/model/claude-fable-5-1.md) | 55.8% | measured |
| 11 | [Muse Wonder](https://benchgap.net/model/muse-wonder.md) | 55.1% | estimated ± 5.7 pp, medium confidence |
| 12 | [Pareto 26.9](https://benchgap.net/model/pareto-26-9.md) | 51.0% | measured |
| 13 | [Pareto 26.10 Preview](https://benchgap.net/model/pareto-26-10-preview.md) | 50.8% | measured |
| 14 | [GPT-6.1 Terra](https://benchgap.net/model/gpt-6-1-terra.md) | 50.3% | measured |
| 15 | [GLM-5.3](https://benchgap.net/model/glm-5-3.md) | 46.9% | estimated ± 5.7 pp, medium confidence |
| 16 | [Ling 3.1 Flash](https://benchgap.net/model/ling-3-1-flash.md) | 46.0% | measured |
| 17 | [DeepSeek V4.1 Flash 1017](https://benchgap.net/model/deepseek-v4-1-flash.md) | 42.3% | measured |
| 18 | [Claude Haiku 5.5](https://benchgap.net/model/claude-haiku-5-5.md) | 39.2% | measured |
| 19 | [MiniMax M3.1 Flash](https://benchgap.net/model/minimax-m3-1-flash.md) | 38.1% | measured |
| 20 | [Grok 4.7](https://benchgap.net/model/grok-4-7.md) | 38.0% | measured |
| 21 | [MiMo-V2.6-Pro](https://benchgap.net/model/mimo-v2-6-pro.md) | 34.9% | measured |
| 22 | [Step 5 Preview](https://benchgap.net/model/step-5-preview.md) | 33.3% | measured |
| 23 | [Mistral Large 4](https://benchgap.net/model/mistral-large-4.md) | 32.8% | estimated ± 5.7 pp, medium confidence |
| 24 | [MiMo-V2.6-Flash](https://benchgap.net/model/mimo-v2-6-flash.md) | 28.8% | measured |
| 25 | [SWE-2](https://benchgap.net/model/swe-2.md) | 27.3% | measured |
| 26 | [Gemini 3.8 Flash](https://benchgap.net/model/gemini-3-8-flash.md) | 19.1% | measured |
| 27 | [GPT-6 Luna](https://benchgap.net/model/gpt-6-luna.md) | 16.9% | estimated ± 5.7 pp, low confidence |
| 28 | [Muse Spark Open](https://benchgap.net/model/muse-spark-open.md) | 11.2% | estimated ± 5.7 pp, low confidence |
| 29 | [Qwen3.8-27B](https://benchgap.net/model/qwen3-8-27b.md) | 7.9% | estimated ± 5.7 pp, low confidence |
| 30 | [Beam 1.0 501B-A23B](https://benchgap.net/model/reflection-beam.md) | 4.3% | estimated ± 5.7 pp, low confidence |
| 31 | [Inkling](https://benchgap.net/model/inkling.md) | 1.5% | estimated ± 5.7 pp, low confidence |

## All leaderboards

### Agentic · terminal

- [AA Terminal-Bench 2.1](https://benchgap.net/b/aa-terminal-bench21/current.md)
- [AA Terminal-Bench 4.0](https://benchgap.net/b/aa-terminal-bench4/current.md)
- [Terminal-Bench 2.0](https://benchgap.net/b/terminal-bench-2/current.md)
- [Terminal-Bench 3.0](https://benchgap.net/b/terminal-bench-3/current.md)
- [Terminal-Bench 4.0](https://benchgap.net/b/terminal-bench-4/current.md)
- [Terminal-Bench Hard](https://benchgap.net/b/terminal-bench-hard/current.md)
- [Terminal-Bench 2.1](https://benchgap.net/b/terminalbench21/current.md)
- [Terminal-Bench 2.1 (Vals AI)](https://benchgap.net/b/vals-terminal-bench21/current.md)

### Agentic · tools

- [AA Agentic Index](https://benchgap.net/b/aa-agentic-index/current.md)
- [AA-AnalystAgent](https://benchgap.net/b/aa-analyst-agent/current.md)
- [AA AutomationBench](https://benchgap.net/b/aa-automation-bench/current.md)
- [CWE-Bench-AA](https://benchgap.net/b/aa-cwe-bench/current.md)
- [AA Cyber Index](https://benchgap.net/b/aa-cyber-index/current.md)
- [DeepsecBench-AA](https://benchgap.net/b/aa-deepsec-bench/current.md)
- [AA EnterpriseOps-Gym](https://benchgap.net/b/aa-enterprise-ops-gym/current.md)
- [GDP.pdf](https://benchgap.net/b/aa-gdp-pdf/current.md)
- [AA Harvey LAB v1.0](https://benchgap.net/b/aa-harvey-lab/current.md)
- [AA ITBench](https://benchgap.net/b/aa-itbench/current.md)
- [AA Tau3 Banking](https://benchgap.net/b/aa-tau3-banking/current.md)
- [Agents' Last Exam](https://benchgap.net/b/agentslastexam/current.md)
- [APEX-Agents](https://benchgap.net/b/apexagents/current.md)
- [APEX-Agents-AA](https://benchgap.net/b/apexagentsaa/current.md)
- [ApprenticeBench](https://benchgap.net/b/apprenticebench/current.md)
- [AutomationBench](https://benchgap.net/b/automationbench/current.md)
- [BFCL v4](https://benchgap.net/b/bfcl-v4/current.md)
- [BrowseComp](https://benchgap.net/b/browsecomp/current.md)
- [Claw-Eval](https://benchgap.net/b/claw-eval/current.md)
- [CWE-bench v1](https://benchgap.net/b/cwe-bench-v1/current.md)
- [CyberGym](https://benchgap.net/b/cybergym/current.md)
- [DeepPlanning](https://benchgap.net/b/deepplanning/current.md)
- [DeepSearchQA](https://benchgap.net/b/deepsearchqa/current.md)
- [DRACO](https://benchgap.net/b/draco/current.md)
- [ExploitGym](https://benchgap.net/b/exploitgym/current.md)
- [Finance Agent v2](https://benchgap.net/b/financeagentv2/current.md)
- [GDPval-AA](https://benchgap.net/b/gdpvalaanormalized/current.md)
- [Gert Labs](https://benchgap.net/b/gertlabs/current.md)
- [Hermes Bench](https://benchgap.net/b/hermesbench/current.md)
- [Hermes Index](https://benchgap.net/b/hermesindex/current.md)
- [HLE w/ tools](https://benchgap.net/b/hlewithtools/current.md)
- [JobBench](https://benchgap.net/b/jobbench/current.md)
- [MazeBench](https://benchgap.net/b/mazebench/current.md)
- [MCP Atlas](https://benchgap.net/b/mcpatlas/current.md)
- [MCP-Tasks](https://benchgap.net/b/mcptasks/current.md)
- [OSWorld 2.1](https://benchgap.net/b/osworld-2-1/current.md)
- [OSWorld 2.1 Offline](https://benchgap.net/b/osworld-2-1-offline/current.md)
- [OSWorld-Verified](https://benchgap.net/b/osworld-verified/current.md)
- [OSWorld 2.0](https://benchgap.net/b/osworld2/current.md)
- [PinchBench](https://benchgap.net/b/pinchbench/current.md)
- [QwenClawBench](https://benchgap.net/b/qwenclawbench/current.md)
- [ResearchClawBench](https://benchgap.net/b/researchclawbench/current.md)
- [SkillsBench](https://benchgap.net/b/skillsbench/current.md)
- [τ²-bench results](https://benchgap.net/b/tau2-bench/current.md)
- [τ³-bench results](https://benchgap.net/b/tau3-bench/current.md)
- [Toolathlon](https://benchgap.net/b/toolathlon/current.md)
- [Toolathlon-Verified](https://benchgap.net/b/toolathlonverified/current.md)
- [VITA-Bench](https://benchgap.net/b/vitabench/current.md)
- [WideResearch](https://benchgap.net/b/wideresearch/current.md)

### Coding

- [AA Coding Index](https://benchgap.net/b/aa-coding-index/current.md)
- [AA LiveCodeBench](https://benchgap.net/b/aa-live-code-bench/current.md)
- [AA-SciCode](https://benchgap.net/b/aa-sci-code/current.md)
- [Bug Hunt Bench](https://benchgap.net/b/bug-hunt-bench/current.md)
- [CursorBench 3.1](https://benchgap.net/b/cursor-bench31/current.md)
- [CursorBench 3.2](https://benchgap.net/b/cursor-bench32/current.md)
- [CursorBench 4.0](https://benchgap.net/b/cursor-bench40/current.md)
- [DeepSWE](https://benchgap.net/b/deepswe/current.md)
- [FrontierCode 1.1 Main](https://benchgap.net/b/frontiercode/current.md)
- [FrontierCode 2.0](https://benchgap.net/b/frontiercode-2/current.md)
- [FrontierCode 1.1 Extended](https://benchgap.net/b/frontiercode11extended/current.md)
- [FrontierSWE v2](https://benchgap.net/b/frontierswev2/current.md)
- [LiveCodeBench](https://benchgap.net/b/livecodebench/current.md)
- [LiveCodeBench Pro](https://benchgap.net/b/livecodebench-pro/current.md)
- [LiveCodeBench v6](https://benchgap.net/b/livecodebench-v6/current.md)
- [NL2Repo](https://benchgap.net/b/nl2repo/current.md)
- [OpenHarmony Bench](https://benchgap.net/b/openharmonybench/current.md)
- [PostTrainBench v1.1](https://benchgap.net/b/posttrainbench-v1-1/current.md)
- [ProgramBench](https://benchgap.net/b/programbench/current.md)
- [React Native Evals](https://benchgap.net/b/reactnativeevals/current.md)
- [SciCode](https://benchgap.net/b/scicode/current.md)
- [SWE-Atlas Codebase QnA](https://benchgap.net/b/swe-atlas-codebase-qna/current.md)
- [SWE-bench Pro](https://benchgap.net/b/swe-bench-pro/current.md)
- [SWE-bench Verified](https://benchgap.net/b/swe-bench-verified/current.md)
- [SWE-Rebench](https://benchgap.net/b/swe-rebench/current.md)
- [SWE-Marathon](https://benchgap.net/b/swemarathon/current.md)
- [Vals LiveCodeBench](https://benchgap.net/b/valslivecodebench/current.md)
- [Vals SWE-bench](https://benchgap.net/b/valsswebench/current.md)
- [Vibe Code Bench](https://benchgap.net/b/vibecodebench/current.md)
- [VulcanBench v3](https://benchgap.net/b/vulcanbench/current.md)

### Math

- [AIME 2024](https://benchgap.net/b/aime2024/current.md)
- [AIME 2025](https://benchgap.net/b/aime2025/current.md)
- [AIME26](https://benchgap.net/b/aime2026/current.md)
- [FrontierMath v2 (Tier 4)](https://benchgap.net/b/frontiermathv2tier4/current.md)
- [FrontierMath v2 (Tiers 1-3)](https://benchgap.net/b/frontiermathv2tiers13/current.md)
- [HMMT Feb 2025](https://benchgap.net/b/hmmtfeb2025/current.md)
- [HMMT Feb 2026](https://benchgap.net/b/hmmtfeb2026/current.md)
- [HMMT Nov 2025](https://benchgap.net/b/hmmtnov2025/current.md)
- [IMOAnswerBench](https://benchgap.net/b/imoanswerbench/current.md)
- [MATH-500](https://benchgap.net/b/math-500/current.md)
- [MMAnswerBench](https://benchgap.net/b/mmanswerbench/current.md)

### Knowledge & reasoning

- [AA-GPQA Diamond](https://benchgap.net/b/aa-gpqa-diamond/current.md)
- [AA-HLE](https://benchgap.net/b/aa-hle/current.md)
- [AA MMLU-Pro](https://benchgap.net/b/aa-mmlu-pro/current.md)
- [ARC-AGI-2](https://benchgap.net/b/arc-agi-2/current.md)
- [ARC-AGI-3 Provider Adapter](https://benchgap.net/b/arc-agi-3-provider-adapter/current.md)
- [ARC-AGI-1](https://benchgap.net/b/arcagi1/current.md)
- [ARC-AGI-3](https://benchgap.net/b/arcagi3/current.md)
- [Artificial Analysis Intelligence Index](https://benchgap.net/b/artificialanalysis/current.md)
- [BBH](https://benchgap.net/b/bbh/current.md)
- [BioMysteryBench (human-difficult)](https://benchgap.net/b/biomysterybenchhumandifficult/current.md)
- [BioMysteryBench (human-solvable)](https://benchgap.net/b/biomysterybenchhumansolvable/current.md)
- [C-Eval](https://benchgap.net/b/ceval/current.md)
- [CritPt](https://benchgap.net/b/critpt/current.md)
- [GPQA](https://benchgap.net/b/gpqa/current.md)
- [GPQA Diamond](https://benchgap.net/b/gpqa-diamond/current.md)
- [GraphWalks BFS 256K–1M](https://benchgap.net/b/graphwalks-bfs-256k-1m/current.md)
- [HealthBench (raw)](https://benchgap.net/b/healthbench/current.md)
- [HealthBench Hard](https://benchgap.net/b/healthbench-hard/current.md)
- [HealthBench (length-adjusted)](https://benchgap.net/b/healthbenchlengthadjusted/current.md)
- [HealthBench Professional](https://benchgap.net/b/healthbenchprofessional/current.md)
- [HealthBench Professional (raw)](https://benchgap.net/b/healthbenchprofessionalraw/current.md)
- [HLE](https://benchgap.net/b/hle/current.md)
- [HLE w/o tools](https://benchgap.net/b/hlenotools/current.md)
- [HLE-Verified](https://benchgap.net/b/hleverified/current.md)
- [JevBench 1.4](https://benchgap.net/b/jev-bench14/current.md)
- [LABBench2](https://benchgap.net/b/labbench2/current.md)
- [MedXpertQA (Text)](https://benchgap.net/b/medxpertqatext/current.md)
- [MMLU](https://benchgap.net/b/mmlu/current.md)
- [MMLU-Pro](https://benchgap.net/b/mmlu-pro/current.md)
- [MMLU-Pro (Arcee)](https://benchgap.net/b/mmlu-pro-arcee/current.md)
- [MMLU-Redux](https://benchgap.net/b/mmluredux/current.md)
- [MMMLU](https://benchgap.net/b/mmmlu/current.md)
- [AA-Omniscience Accuracy](https://benchgap.net/b/omniscienceaccuracy/current.md)
- [SimpleQA](https://benchgap.net/b/simpleqa/current.md)
- [STEMBench](https://benchgap.net/b/stembench/current.md)
- [SuperGPQA](https://benchgap.net/b/supergpqa/current.md)
- [Vals GPQA Diamond](https://benchgap.net/b/vals-gpqa-diamond/current.md)
- [Vals MMLU-Pro](https://benchgap.net/b/vals-mmlu-pro/current.md)

### Instruction following

- [AA-IFBench](https://benchgap.net/b/aa-if-bench/current.md)
- [IFBench](https://benchgap.net/b/ifbench/current.md)
- [IFEval](https://benchgap.net/b/ifeval/current.md)

### Multilingual

- [INCLUDE](https://benchgap.net/b/include/current.md)
- [MAXIFE](https://benchgap.net/b/maxife/current.md)
- [MMLU-ProX](https://benchgap.net/b/mmluprox/current.md)
- [NOVA-63](https://benchgap.net/b/nova63/current.md)
- [PolyMath](https://benchgap.net/b/polymath/current.md)
- [SWE Multilingual](https://benchgap.net/b/swe-bench-multilingual/current.md)

### Vision & documents

- [AA-MMMU-Pro](https://benchgap.net/b/aa-mmmu-pro/current.md)
- [AI2D_TEST](https://benchgap.net/b/ai2dtest/current.md)
- [BabyVision](https://benchgap.net/b/babyvision/current.md)
- [CC-OCR](https://benchgap.net/b/ccocr/current.md)
- [Chartography (tools)](https://benchgap.net/b/chartographywithtools/current.md)
- [CharXiv](https://benchgap.net/b/charxiv/current.md)
- [CharXiv w/o tools](https://benchgap.net/b/charxivnotools/current.md)
- [CountBench](https://benchgap.net/b/countbench/current.md)
- [DynaMath](https://benchgap.net/b/dynamath/current.md)
- [ERQA](https://benchgap.net/b/erqa/current.md)
- [LVBench](https://benchgap.net/b/lvbench/current.md)
- [MathVision](https://benchgap.net/b/mathvision/current.md)
- [MathVision w/ Python](https://benchgap.net/b/mathvisionpython/current.md)
- [MedXpertQA (MM)](https://benchgap.net/b/medxpertqamm/current.md)
- [MLVU (M-Avg)](https://benchgap.net/b/mlvuavg/current.md)
- [MMMU](https://benchgap.net/b/mmmu/current.md)
- [MMMU-Pro](https://benchgap.net/b/mmmu-pro/current.md)
- [MMMU-Pro w/ Python](https://benchgap.net/b/mmmupropython/current.md)
- [MMVU](https://benchgap.net/b/mmvu/current.md)
- [OCRBench V2](https://benchgap.net/b/ocrbenchv2/current.md)
- [OfficeQA Pro](https://benchgap.net/b/officeqapro/current.md)
- [OmniDocBench 1.5](https://benchgap.net/b/omnidocbench15/current.md)
- [ParseBench](https://benchgap.net/b/parsebench/current.md)
- [RealWorldQA](https://benchgap.net/b/realworldqa/current.md)
- [RefCOCO (avg)](https://benchgap.net/b/refcocoavg/current.md)
- [ScreenSpot Pro](https://benchgap.net/b/screenspot-pro/current.md)
- [SimpleVQA](https://benchgap.net/b/simplevqa/current.md)
- [SWE-bench Multimodal](https://benchgap.net/b/swe-bench-multimodal/current.md)
- [Video-MME (w/o subtitle)](https://benchgap.net/b/videommenosub/current.md)
- [Video-MME (with subtitle)](https://benchgap.net/b/videommewithsub/current.md)
- [VideoMMMU](https://benchgap.net/b/videommmu/current.md)
- [V*](https://benchgap.net/b/vstar/current.md)
- [We-Math](https://benchgap.net/b/wemath/current.md)
- [ZeroBench](https://benchgap.net/b/zerobench/current.md)

### Long context

- [AA-LCR](https://benchgap.net/b/lcr/current.md)
- [LongBench v2](https://benchgap.net/b/longbench-v2/current.md)
- [MMLongBench-Doc](https://benchgap.net/b/mmlongbenchdoc/current.md)
- [MRCRv2](https://benchgap.net/b/mrcrv2/current.md)

### General

- [HLE-MultiChoice (Decision Index)](https://benchgap.net/b/aa-decision-hle-multi-choice/current.md)
- [Artificial Analysis Decision Index](https://benchgap.net/b/aa-decision-index/current.md)
- [When2Call (Decision Index)](https://benchgap.net/b/aa-decision-when2-call/current.md)
- [WildGuardTest (Decision Index)](https://benchgap.net/b/aa-decision-wild-guard-test/current.md)
- [Decision Index Full Score](https://benchgap.net/b/decisionindex/current.md)
- [ExploitBench](https://benchgap.net/b/exploitbench/current.md)
- [Decision Index Arts](https://benchgap.net/b/jevdecisionindex03arts/current.md)
- [Decision Index Knowledge](https://benchgap.net/b/jevdecisionindex03knowledge/current.md)
- [Decision Index Language](https://benchgap.net/b/jevdecisionindex03language/current.md)
- [Decision Index Retrieval](https://benchgap.net/b/jevdecisionindex03retrieval/current.md)
- [Decision Index Tools](https://benchgap.net/b/jevdecisionindex03tools/current.md)
- [SEC-Bench Pro](https://benchgap.net/b/secbenchpro/current.md)