Anthropic · model
Claude Fable 5.2 benchmark scores
As of 2026-11-18, Claude Fable 5.2 (Anthropic) has measured scores on 47 benchmarks and estimated scores on 98 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 95.3% | estimated ± 2.7 pp, medium confidence |
| AA-HLE | 74.3% | measured |
| AA-SciCode | 67.7% | measured |
| AIME26 | 98.2% | estimated ± 1.4 pp, medium confidence |
| Artificial Analysis Intelligence Index | 64.6% | measured |
| CritPt | 35.4% | measured |
| GDPval-AA | 70.6% | estimated ± 1.7 pp, low confidence |
| GPQA Diamond | 96.6% | estimated ± 1.8 pp, medium confidence |
| HLE | 74.2% | measured |
| AA-LCR | 83.3% | measured |
| LiveCodeBench v6 | 100.0% | estimated ± 2.7 pp, medium confidence |
| AA-Omniscience Accuracy | 79.1% | measured |
| SciCode | 67.0% | estimated ± 3.1 pp, medium confidence |
| Terminal-Bench 2.1 | 89.1% | estimated ± 5.4 pp, low confidence |
| BrowseComp | 91.0% | estimated ± 2.7 pp, medium confidence |
| HLE w/ tools | 64.2% | estimated ± 6.8 pp, low confidence |
| LongBench v2 | 68.6% | estimated ± 5.5 pp, medium confidence |
| AA Coding Index | 86.4% | estimated ± 1.2 pp, medium confidence |
| AA-MMMU-Pro | 100.0% | estimated ± 2.3 pp, medium confidence |
| APEX-Agents | 51.6% | estimated ± 7.4 pp, low confidence |
| BioMysteryBench (human-difficult) | 50.2% | estimated ± 6.8 pp, low confidence |
| SWE-bench Verified | 100.0% | estimated ± 3.6 pp, medium confidence |
| AutomationBench | 51.4% | measured |
| CyberGym | 87.4% | estimated ± 4.6 pp, medium confidence |
| JobBench | 69.2% | estimated ± 11.3 pp, low confidence |
| SWE-bench Pro | 93.0% | measured |
| MMLU-Pro | 86.2% | estimated ± 2.5 pp, low confidence |
| FailureSensorIQ | 68.6% | estimated ± 1.7 pp, high confidence |
| GPQA | 96.5% | estimated ± 0.9 pp, high confidence |
| Vals GPQA Diamond | 96.5% | estimated ± 2.9 pp, medium confidence |
| Vals LiveCodeBench | 94.6% | estimated ± 3.0 pp, medium confidence |
| Vals MMLU-Pro | 92.8% | estimated ± 1.8 pp, medium confidence |
| MMMLU | 95.5% | estimated ± 2.6 pp, low confidence |
| Chess Puzzles | 79.0% | measured |
| AA MMLU-Pro | 93.2% | estimated ± 0.8 pp, medium confidence |
| AA-AnalystAgent | 58.1% | estimated ± 2.7 pp, low confidence |
| AA Tau3 Banking | 50.0% | estimated ± 3.3 pp, low confidence |
| ApprenticeBench | 60.8% | estimated ± 12.5 pp, low confidence |
| ARC-AGI-1 | 97.8% | estimated ± 3.4 pp, high confidence |
| ARC-AGI-2 | 92.7% | estimated ± 9.4 pp, medium confidence |
| BabaIsBench | 100.0% | estimated ± 1.2 pp, medium confidence |
| CursorBench 3.1 | 87.6% | estimated ± 3.4 pp, low confidence |
| CursorBench 3.2 | 77.2% | estimated ± 2.0 pp, medium confidence |
| EBR-bench | 90.5% | measured |
| Terminal-Bench 3.0 | 50.4% | estimated ± 3.2 pp, medium confidence |
| FrontierCode 1.1 Main | 52.1% | estimated ± 3.5 pp, low confidence |
| FrontierMath v2 (Tier 4) | 100.0% | measured |
| FrontierMath v2 (Tiers 1-3) | 95.1% | measured |
| FrontierSWE v2 | 69.9% | measured |
| Furniture Assembly | 90.0% | measured |
| MaxxBench | 76.6% | estimated ± 4.3 pp, low confidence |
| MazeBench | 21.0% | measured |
| Mystery Game Puzzles | 90.0% | measured |
| NYT Connections | 97.3% | measured |
| OfficeQA Pro | 72.9% | measured |
| ParseBench-1 | 86.6% | estimated ± 4.6 pp, low confidence |
| Terminal-Bench 4.1 | 65.8% | measured |
| Terminal-Bench Hard | 70.7% | estimated ± 2.8 pp, low confidence |
| Vals SWE-bench | 100.0% | estimated ± 3.7 pp, medium confidence |
| Terminal-Bench 2.1 (Vals AI) | 82.6% | estimated ± 6.9 pp, low confidence |
| VulcanBench v3 | 87.6% | estimated ± 3.9 pp, low confidence |
| CWE-Bench-AA | 62.9% | estimated ± 4.1 pp, medium confidence |
| AA Cyber Index | 30.5% | estimated ± 5.3 pp, low confidence |
| DeepsecBench-AA | 33.4% | estimated ± 6.1 pp, low confidence |
| Harvey LAB-AA v1.1 | 7.2% | measured |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 67.7% | measured |
| HealthBench Professional (raw, Artificial Analysis) | 71.1% | measured |
| BenchCAD Vision2Code (no tools) | 0.8% | estimated ± 0.1 pp, low confidence |
| BenchCAD Vision2Code (tools) | 1.0% | estimated ± 0.0 pp, low confidence |
| Bug Hunt Bench | 58.0% | measured |
| Chartography (no tools) | 73.4% | measured |
| Chartography (tools) | 91.6% | measured |
| CursorBench 4.0 | 60.0% | measured |
| CWE-bench v1 | 68.9% | estimated ± 5.3 pp, medium confidence |
| DeepSWE | 74.3% | measured |
| DRACO | 87.7% | measured |
| Epoch Automation Reports | 65.7% | estimated ± 3.0 pp, low confidence |
| FrontierCode 2.0 | 50.2% | measured |
| FrontierPhysics | 50.5% | estimated ± 3.2 pp, medium confidence |
| HLE w/o tools | 73.2% | measured |
| OfficeQA | 83.4% | measured |
| OSWorld 2.1 | 78.6% | estimated ± 6.1 pp, medium confidence |
| OSWorld 2.1 Offline | 89.3% | measured |
| PostTrainBench v1.1 | 49.3% | estimated ± 1.7 pp, low confidence |
| ProgramBench | 85.6% | measured |
| SpireClimb | 15.3% | measured |
| StealthMark | 74.7% | estimated ± 6.5 pp, low confidence |
| SWE Multilingual | 96.6% | measured |
| SWE-bench Multimodal | 74.1% | measured |
| Terminal-Bench 4.0 | 67.6% | measured |
| Terminal-Bench-Science 0.1 | 74.6% | measured |
| Toolathlon-Verified | 78.4% | estimated ± 1.2 pp, medium confidence |
| VulcanBench Frontier v4 | 93.9% | measured |
| AA AutomationBench | 76.4% | measured |
| GDP.pdf | 24.2% | measured |
| AA Terminal-Bench 4.0 | 59.6% | measured |
| ARC-AGI-3 | 65.7% | measured |
| ARC-AGI-3 Provider Adapter | 94.9% | measured |
| CWE-bench | 75.0% | measured |
| MKA | 56.6% | measured |
| SkillsBench official leaderboard (BenchFlow) | 60.3% | estimated ± 6.1 pp, low confidence |
| BinaryAudit | 35.8% | estimated ± 12.1 pp, low confidence |
| OTelBench | 22.5% | estimated ± 6.6 pp, low confidence |
| FACTS Multimodal | 54.2% | estimated ± 4.0 pp, high confidence |
| Hermes Bench | 71.9% | estimated ± 1.8 pp, medium confidence |
| Hermes Index | 38.2% | estimated ± 2.7 pp, medium confidence |
| CharXiv | 91.7% | estimated ± 5.1 pp, low confidence |
| ExploitBench | 100.0% | estimated ± 6.4 pp, low confidence |
| ExploitGym | 43.5% | estimated ± 6.4 pp, medium confidence |
| HMMT Feb 2026 | 100.0% | estimated ± 3.1 pp, medium confidence |
| HMMT Nov 2025 | 96.2% | estimated ± 1.5 pp, low confidence |
| INCLUDE | 91.1% | estimated ± 2.0 pp, low confidence |
| MCP Atlas | 85.0% | estimated ± 7.6 pp, low confidence |
| MMAnswerBench | 98.1% | estimated ± 2.4 pp, low confidence |
| NL2Repo | 62.7% | estimated ± 6.0 pp, medium confidence |
| SuperGPQA | 100.0% | estimated ± 7.2 pp, low confidence |
| HealthBench Hard | 34.4% | estimated ± 8.6 pp, low confidence |
| LiveCodeBench Pro | 100.0% | estimated ± 6.0 pp, low confidence |
| React Native Evals | 84.2% | estimated ± 4.7 pp, low confidence |
| PostTrainBench | 42.2% | estimated ± 2.4 pp, medium confidence |
| DAYJOB | 19.1% | estimated ± 2.7 pp, medium confidence |
| DAYJOB Healthcare | 18.2% | estimated ± 4.0 pp, medium confidence |
| Finance Agent v2 | 62.3% | estimated ± 3.8 pp, low confidence |
| BioMysteryBench (human-solvable) | 89.5% | estimated ± 0.6 pp, low confidence |
| FrontierCode 1.1 Extended | 64.0% | estimated ± 3.0 pp, low confidence |
| HealthBench (raw) | 74.8% | estimated ± 6.0 pp, low confidence |
| HealthBench (length-adjusted) | 58.4% | estimated ± 0.9 pp, low confidence |
| HealthBench Professional | 63.7% | estimated ± 2.1 pp, medium confidence |
| HealthBench Professional (raw) | 66.4% | estimated ± 2.3 pp, medium confidence |
| HLE-Verified | 90.9% | estimated ± 7.1 pp, low confidence |
| LABBench2 | 90.7% | estimated ± 2.5 pp, low confidence |
| ScarfBench | 68.9% | estimated ± 3.8 pp, medium confidence |
| BlenderWorld (text-only) | 86.5% | estimated ± 2.8 pp, low confidence |
| Cookie-Bench | 90.8% | estimated ± 1.3 pp, low confidence |
| STEMBench | 2.8% | estimated ± 0.2 pp, low confidence |
| Agents' Last Exam | 52.3% | estimated ± 10.7 pp, low confidence |
| OpenHarmony Bench v1.0 | 58.9% | estimated ± 2.4 pp, medium confidence |
| OpenHarmony Bench v1.2 | 63.8% | estimated ± 11.0 pp, low confidence |
| SEC-Bench Pro | 76.9% | estimated ± 6.9 pp, low confidence |
| MRCRv2 | 85.9% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 61.5% | estimated ± 2.6 pp, low confidence |
| SWE-Marathon | 68.3% | estimated ± 2.1 pp, low confidence |
| BioBasicsBench | 69.0% | estimated ± 3.5 pp, low confidence |
| ExploitBench Internal Port | 57.0% | estimated ± 1.4 pp, low confidence |
| SWE-Atlas Codebase QnA | 61.5% | estimated ± 4.8 pp, high confidence |