Anthropic · model
Claude Opus 4.5 Thinking benchmark scores
As of 2026-10-07, Claude Opus 4.5 Thinking (Anthropic) has measured scores on 12 benchmarks and estimated scores on 52 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 86.6% | measured |
| AA-HLE | 30.1% | measured |
| AA-SciCode | 48.7% | estimated ± 3.8 pp, high confidence |
| Artificial Analysis Intelligence Index | 29.1% | measured |
| CritPt | 4.6% | measured |
| GDPval-AA | 22.3% | estimated ± 12.5 pp, low confidence |
| GPQA Diamond | 87.6% | estimated ± 1.4 pp, medium confidence |
| HLE | 44.0% | estimated ± 6.3 pp, medium confidence |
| IFBench | 60.3% | estimated ± 6.9 pp, medium confidence |
| AA-LCR | 77.3% | measured |
| AA-Omniscience Accuracy | 46.6% | measured |
| IFEval | 92.5% | estimated ± 1.3 pp, high confidence |
| LongBench v2 | 61.5% | estimated ± 5.9 pp, medium confidence |
| τ²-bench results | 89.5% | measured |
| AA Coding Index | 49.9% | estimated ± 6.8 pp, medium confidence |
| AA-MMMU-Pro | 74.0% | measured |
| APEX-Agents-AA | 25.3% | estimated ± 11.4 pp, low confidence |
| BioMysteryBench (human-difficult) | 44.1% | estimated ± 5.9 pp, low confidence |
| SWE-bench Verified | 75.5% | estimated ± 4.4 pp, high confidence |
| BFCL v4 | 64.9% | estimated ± 9.8 pp, low confidence |
| AA Agentic Index | 22.5% | estimated ± 12.3 pp, low confidence |
| MMLU-Pro | 85.7% | estimated ± 3.5 pp, high confidence |
| GPQA | 88.9% | estimated ± 1.6 pp, high confidence |
| AA-IFBench | 58.0% | measured |
| AA EnterpriseOps-Gym | 37.0% | estimated ± 3.6 pp, medium confidence |
| ARC-AGI-1 | 80.0% | measured |
| ARC-AGI-2 | 37.6% | measured |
| FrontierCode 1.1 Main | 8.5% | estimated ± 4.3 pp, low confidence |
| OfficeQA Pro | 50.3% | estimated ± 7.2 pp, low confidence |
| Vals GPQA Diamond | 87.3% | estimated ± 2.2 pp, high confidence |
| Vals MMLU-Pro | 86.5% | estimated ± 1.8 pp, high confidence |
| Vals SWE-bench | 72.6% | estimated ± 6.3 pp, medium confidence |
| AA ITBench | 49.3% | estimated ± 3.6 pp, medium confidence |
| HLE w/o tools | 27.2% | estimated ± 5.7 pp, medium confidence |
| PostTrainBench v1.1 | 19.9% | estimated ± 10.5 pp, low confidence |
| Vibe Code Bench | 20.6% | measured |
| CharXiv | 79.2% | estimated ± 7.5 pp, low confidence |
| CharXiv w/o tools | 78.9% | estimated ± 2.4 pp, high confidence |
| C-Eval | 93.9% | estimated ± 1.0 pp, low confidence |
| MathVision | 83.5% | estimated ± 4.9 pp, high confidence |
| MCP-Tasks | 69.8% | estimated ± 8.2 pp, low confidence |
| MMLU-Redux | 97.2% | estimated ± 1.1 pp, low confidence |
| MMMU-Pro | 76.5% | estimated ± 4.4 pp, medium confidence |
| ScreenSpot Pro | 72.7% | estimated ± 10.7 pp, low confidence |
| SuperGPQA | 84.3% | estimated ± 7.5 pp, low confidence |
| ERQA | 59.7% | estimated ± 3.9 pp, high confidence |
| HealthBench Hard | 25.5% | estimated ± 7.7 pp, low confidence |
| MedXpertQA (MM) | 69.7% | estimated ± 6.5 pp, low confidence |
| MMLU-Pro (Arcee) | 94.4% | estimated ± 4.0 pp, low confidence |
| React Native Evals | 69.0% | estimated ± 3.0 pp, low confidence |
| SWE-Rebench | 34.7% | estimated ± 6.5 pp, low confidence |
| ARC-AGI-3 | 1.7% | estimated ± 3.4 pp, high confidence |
| BioMysteryBench (human-solvable) | 81.8% | estimated ± 0.6 pp, low confidence |
| HealthBench (raw) | 39.3% | estimated ± 6.4 pp, low confidence |
| HealthBench (length-adjusted) | 51.1% | estimated ± 2.5 pp, low confidence |
| HealthBench Professional | 55.1% | estimated ± 4.6 pp, medium confidence |
| HealthBench Professional (raw) | 50.4% | estimated ± 5.0 pp, low confidence |
| HLE-Verified | 35.3% | estimated ± 0.7 pp, low confidence |
| LABBench2 | 51.6% | estimated ± 1.3 pp, low confidence |
| MMMU | 82.2% | estimated ± 0.9 pp, high confidence |
| MMLU | 94.5% | estimated ± 1.1 pp, low confidence |
| MMMU-Pro w/ Python | 78.0% | estimated ± 1.6 pp, high confidence |
| RealWorldQA | 84.4% | estimated ± 1.2 pp, low confidence |
| OmniDocBench 1.5 | 67.2% | estimated ± 6.4 pp, low confidence |