# GPT-OSS 20B benchmark scores

OpenAI · model · https://benchgap.net/model/gpt-oss-20b · data retrieved 2026-11-22

As of 2026-11-22, GPT-OSS 20B (OpenAI) has measured scores on 19 benchmarks and estimated scores on 80 more.

| Benchmark | Score | Source |
|---|---|---|
| [AA-GPQA Diamond](https://benchgap.net/b/aa-gpqa-diamond/current.md) | 68.8% | measured |
| [AA-HLE](https://benchgap.net/b/aa-hle/current.md) | 11.0% | measured |
| [AA-SciCode](https://benchgap.net/b/aa-sci-code/current.md) | 38.9% | measured |
| [AIME26](https://benchgap.net/b/aime2026/current.md) | 90.9% | estimated ± 4.3 pp, medium confidence |
| [Artificial Analysis Intelligence Index](https://benchgap.net/b/artificialanalysis/current.md) | 9.8% | estimated ± 3.2 pp, medium confidence |
| [CritPt](https://benchgap.net/b/critpt/current.md) | 1.4% | measured |
| [GDPval-AA](https://benchgap.net/b/gdpvalaanormalized/current.md) | 0.0% | measured |
| [GPQA Diamond](https://benchgap.net/b/gpqa-diamond/current.md) | 71.7% | estimated ± 1.2 pp, high confidence |
| [HLE](https://benchgap.net/b/hle/current.md) | 10.9% | measured |
| [IFBench](https://benchgap.net/b/ifbench/current.md) | 63.7% | estimated ± 6.6 pp, medium confidence |
| [AA-LCR](https://benchgap.net/b/lcr/current.md) | 34.7% | measured |
| [LiveCodeBench v6](https://benchgap.net/b/livecodebench-v6/current.md) | 71.9% | estimated ± 4.7 pp, high confidence |
| [AA-Omniscience Accuracy](https://benchgap.net/b/omniscienceaccuracy/current.md) | 16.5% | estimated ± 2.4 pp, high confidence |
| [SciCode](https://benchgap.net/b/scicode/current.md) | 39.1% | estimated ± 3.1 pp, high confidence |
| [BrowseComp](https://benchgap.net/b/browsecomp/current.md) | 8.5% | estimated ± 4.2 pp, medium confidence |
| [HLE w/ tools](https://benchgap.net/b/hlewithtools/current.md) | 26.8% | estimated ± 4.1 pp, low confidence |
| [IFEval](https://benchgap.net/b/ifeval/current.md) | 92.5% | estimated ± 2.8 pp, high confidence |
| [LongBench v2](https://benchgap.net/b/longbench-v2/current.md) | 51.3% | estimated ± 5.5 pp, medium confidence |
| [τ²-bench results](https://benchgap.net/b/tau2-bench/current.md) | 60.2% | measured |
| [AA Coding Index](https://benchgap.net/b/aa-coding-index/current.md) | 20.7% | measured |
| [APEX-Agents](https://benchgap.net/b/apexagents/current.md) | 0.0% | estimated ± 9.4 pp, low confidence |
| [APEX-Agents-AA](https://benchgap.net/b/apexagentsaa/current.md) | 0.7% | measured |
| [BioMysteryBench (human-difficult)](https://benchgap.net/b/biomysterybenchhumandifficult/current.md) | 20.9% | estimated ± 6.8 pp, low confidence |
| [DeepSearchQA](https://benchgap.net/b/deepsearchqa/current.md) | 26.7% | estimated ± 4.5 pp, medium confidence |
| [SWE-bench Verified](https://benchgap.net/b/swe-bench-verified/current.md) | 60.7% | measured |
| [BFCL v4](https://benchgap.net/b/bfcl-v4/current.md) | 53.9% | estimated ± 2.7 pp, high confidence |
| [CyberGym](https://benchgap.net/b/cybergym/current.md) | 46.6% | estimated ± 5.3 pp, low confidence |
| [JobBench](https://benchgap.net/b/jobbench/current.md) | 0.0% | estimated ± 8.2 pp, low confidence |
| [SWE-bench Pro](https://benchgap.net/b/swe-bench-pro/current.md) | 37.1% | estimated ± 4.5 pp, high confidence |
| [WideResearch](https://benchgap.net/b/wideresearch/current.md) | 0.0% | estimated ± 7.8 pp, low confidence |
| [AA Agentic Index](https://benchgap.net/b/aa-agentic-index/current.md) | 1.4% | measured |
| [MMLU-Pro](https://benchgap.net/b/mmlu-pro/current.md) | 72.3% | estimated ± 4.0 pp, high confidence |
| [AA-IFBench](https://benchgap.net/b/aa-if-bench/current.md) | 65.1% | measured |
| [AIME 2024](https://benchgap.net/b/aime2024/current.md) | 92.1% | measured |
| [BBH](https://benchgap.net/b/bbh/current.md) | 90.2% | estimated ± 2.4 pp, low confidence |
| [FailureSensorIQ](https://benchgap.net/b/failuresensoriq/current.md) | 43.5% | measured |
| [GPQA](https://benchgap.net/b/gpqa/current.md) | 71.5% | measured |
| [MMLU](https://benchgap.net/b/mmlu/current.md) | 85.3% | measured |
| [Vals GPQA Diamond](https://benchgap.net/b/vals-gpqa-diamond/current.md) | 67.9% | estimated ± 3.2 pp, high confidence |
| [Vals LiveCodeBench](https://benchgap.net/b/valslivecodebench/current.md) | 75.8% | estimated ± 5.7 pp, low confidence |
| [Vals MMLU-Pro](https://benchgap.net/b/vals-mmlu-pro/current.md) | 79.6% | estimated ± 2.3 pp, high confidence |
| [MATH-500](https://benchgap.net/b/math-500/current.md) | 97.9% | estimated ± 1.2 pp, medium confidence |
| [MMMLU](https://benchgap.net/b/mmmlu/current.md) | 86.8% | estimated ± 0.7 pp, medium confidence |
| [AIME 2025](https://benchgap.net/b/aime2025/current.md) | 91.7% | measured |
| [AA MMLU-Pro](https://benchgap.net/b/aa-mmlu-pro/current.md) | 85.3% | estimated ± 1.2 pp, low confidence |
| [Gert Labs](https://benchgap.net/b/gertlabs/current.md) | 28.4% | estimated ± 9.3 pp, low confidence |
| [AA Harvey LAB v1.0](https://benchgap.net/b/aa-harvey-lab/current.md) | 79.4% | estimated ± 1.2 pp, low confidence |
| [AA Tau3 Banking](https://benchgap.net/b/aa-tau3-banking/current.md) | 16.8% | estimated ± 3.7 pp, medium confidence |
| [ARC-AGI-1](https://benchgap.net/b/arcagi1/current.md) | 24.7% | estimated ± 6.7 pp, low confidence |
| [ARC-AGI-2](https://benchgap.net/b/arc-agi-2/current.md) | 0.0% | estimated ± 10.9 pp, low confidence |
| [CursorBench 3.2](https://benchgap.net/b/cursor-bench32/current.md) | 36.0% | estimated ± 5.3 pp, low confidence |
| [FrontierCode 1.1 Main](https://benchgap.net/b/frontiercode/current.md) | 0.0% | estimated ± 6.6 pp, low confidence |
| [FrontierSWE v2](https://benchgap.net/b/frontierswev2/current.md) | 4.6% | estimated ± 14.2 pp, low confidence |
| [MazeBench](https://benchgap.net/b/mazebench/current.md) | 0.4% | estimated ± 5.5 pp, low confidence |
| [OSWorld-Verified](https://benchgap.net/b/osworld-verified/current.md) | 55.4% | estimated ± 8.4 pp, low confidence |
| [Vals SWE-bench](https://benchgap.net/b/valsswebench/current.md) | 44.6% | estimated ± 5.5 pp, medium confidence |
| [CWE-Bench-AA](https://benchgap.net/b/aa-cwe-bench/current.md) | 0.0% | estimated ± 10.1 pp, low confidence |
| [AA Cyber Index](https://benchgap.net/b/aa-cyber-index/current.md) | 1.0% | estimated ± 9.2 pp, low confidence |
| [DeepsecBench-AA](https://benchgap.net/b/aa-deepsec-bench/current.md) | 0.0% | estimated ± 8.9 pp, low confidence |
| [AA ITBench](https://benchgap.net/b/aa-itbench/current.md) | 33.3% | estimated ± 4.0 pp, medium confidence |
| [Bug Hunt Bench](https://benchgap.net/b/bug-hunt-bench/current.md) | 13.9% | estimated ± 10.5 pp, low confidence |
| [CursorBench 4.0](https://benchgap.net/b/cursor-bench40/current.md) | 25.6% | estimated ± 4.9 pp, medium confidence |
| [CWE-bench v1](https://benchgap.net/b/cwe-bench%23cwe-bench-v1/current.md) | 0.0% | estimated ± 6.4 pp, low confidence |
| [DeepSWE](https://benchgap.net/b/deepswe/current.md) | 28.7% | estimated ± 7.5 pp, low confidence |
| [DRACO](https://benchgap.net/b/draco/current.md) | 0.0% | estimated ± 2.0 pp, low confidence |
| [FrontierCode 2.0](https://benchgap.net/b/frontiercode-2/current.md) | 26.1% | estimated ± 9.3 pp, low confidence |
| [HLE w/o tools](https://benchgap.net/b/hlenotools/current.md) | 11.5% | estimated ± 5.3 pp, medium confidence |
| [PostTrainBench v1.1](https://benchgap.net/b/posttrainbench%23posttrainbench-v1-1/current.md) | 25.3% | estimated ± 7.4 pp, low confidence |
| [Toolathlon-Verified](https://benchgap.net/b/toolathlonverified/current.md) | 5.5% | estimated ± 2.3 pp, low confidence |
| [VulcanBench Frontier v4](https://benchgap.net/b/vulcanbench/current.md) | 68.5% | estimated ± 6.8 pp, low confidence |
| [GDP.pdf](https://benchgap.net/b/aa-gdp-pdf/current.md) | 17.0% | estimated ± 4.6 pp, medium confidence |
| [ARC-AGI-3](https://benchgap.net/b/arcagi3/current.md) | 1.3% | estimated ± 9.8 pp, low confidence |
| [ARC-AGI-3 Provider Adapter](https://benchgap.net/b/arc-agi-3-provider-adapter/current.md) | 2.7% | estimated ± 7.3 pp, low confidence |
| [Vibe Code Bench](https://benchgap.net/b/vibecodebench/current.md) | 27.7% | estimated ± 6.5 pp, low confidence |
| [Hermes Bench](https://benchgap.net/b/hermesbench/current.md) | 0.0% | estimated ± 2.9 pp, medium confidence |
| [Hermes Index](https://benchgap.net/b/hermesindex/current.md) | 0.0% | estimated ± 8.7 pp, low confidence |
| [C-Eval](https://benchgap.net/b/ceval/current.md) | 92.2% | estimated ± 1.1 pp, medium confidence |
| [HMMT Feb 2025](https://benchgap.net/b/hmmtfeb2025/current.md) | 94.9% | estimated ± 2.9 pp, high confidence |
| [HMMT Feb 2026](https://benchgap.net/b/hmmtfeb2026/current.md) | 76.2% | estimated ± 11.0 pp, low confidence |
| [MCP Atlas](https://benchgap.net/b/mcpatlas/current.md) | 64.6% | estimated ± 7.8 pp, low confidence |
| [MMLU-Redux](https://benchgap.net/b/mmluredux/current.md) | 90.9% | estimated ± 1.7 pp, high confidence |
| [NL2Repo](https://benchgap.net/b/nl2repo/current.md) | 12.1% | estimated ± 4.7 pp, medium confidence |
| [QwenClawBench](https://benchgap.net/b/qwenclawbench/current.md) | 55.9% | estimated ± 6.0 pp, low confidence |
| [SuperGPQA](https://benchgap.net/b/supergpqa/current.md) | 52.5% | estimated ± 7.2 pp, medium confidence |
| [Toolathlon](https://benchgap.net/b/toolathlon/current.md) | 27.0% | estimated ± 6.3 pp, low confidence |
| [AA LiveCodeBench](https://benchgap.net/b/aa-live-code-bench/current.md) | 84.0% | estimated ± 4.7 pp, low confidence |
| [HealthBench Hard](https://benchgap.net/b/healthbench-hard/current.md) | 12.3% | estimated ± 8.6 pp, low confidence |
| [MMLU-Pro (Arcee)](https://benchgap.net/b/mmlu-pro-arcee/current.md) | 82.4% | estimated ± 4.0 pp, medium confidence |
| [React Native Evals](https://benchgap.net/b/reactnativeevals/current.md) | 71.0% | measured |
| [ResearchClawBench](https://benchgap.net/b/researchclawbench/current.md) | 2.8% | estimated ± 1.9 pp, medium confidence |
| [SWE-Rebench](https://benchgap.net/b/swe-rebench/current.md) | 57.0% | estimated ± 3.6 pp, low confidence |
| [BioMysteryBench (human-solvable)](https://benchgap.net/b/biomysterybenchhumansolvable/current.md) | 81.7% | estimated ± 0.6 pp, low confidence |
| [ScarfBench](https://benchgap.net/b/scarfbench/current.md) | 24.4% | estimated ± 9.1 pp, low confidence |
| [SimpleQA](https://benchgap.net/b/simpleqa/current.md) | 22.9% | estimated ± 7.8 pp, medium confidence |
| [LiveCodeBench](https://benchgap.net/b/livecodebench/current.md) | 66.5% | estimated ± 14.6 pp, low confidence |
| [OpenHarmony Bench v1.0](https://benchgap.net/b/openharmonybench%23openharmonybench-v1-0/current.md) | 51.0% | estimated ± 2.4 pp, medium confidence |
| [MRCRv2](https://benchgap.net/b/mrcrv2/current.md) | 22.8% | estimated ± 11.3 pp, low confidence |
| [MMLongBench-Doc](https://benchgap.net/b/mmlongbenchdoc/current.md) | 57.3% | estimated ± 2.6 pp, low confidence |
| [BioBasicsBench](https://benchgap.net/b/biobasicsbench/current.md) | 33.9% | estimated ± 10.7 pp, low confidence |
