xAI · model
Grok 4.5 benchmark scores
As of 2026-11-22, Grok 4.5 (xAI) has measured scores on 39 benchmarks and estimated scores on 138 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 93.1% | measured |
| AA-HLE | 42.7% | measured |
| AA-SciCode | 55.0% | measured |
| AIME26 | 96.0% | estimated ± 1.2 pp, medium confidence |
| Artificial Analysis Intelligence Index | 38.8% | measured |
| CritPt | 15.4% | measured |
| GDPval-AA | 44.2% | measured |
| GPQA Diamond | 89.9% | estimated ± 1.4 pp, high confidence |
| HLE | 54.3% | estimated ± 2.2 pp, high confidence |
| AA-LCR | 79.3% | measured |
| LiveCodeBench v6 | 89.8% | estimated ± 2.2 pp, low confidence |
| AA-Omniscience Accuracy | 51.6% | measured |
| SciCode | 51.5% | estimated ± 2.9 pp, medium confidence |
| Terminal-Bench 2.1 | 83.3% | measured |
| BrowseComp | 87.2% | estimated ± 4.7 pp, high confidence |
| HLE w/ tools | 55.2% | estimated ± 5.3 pp, medium confidence |
| LongBench v2 | 65.1% | estimated ± 5.5 pp, medium confidence |
| AA Coding Index | 72.5% | measured |
| AA-MMMU-Pro | 80.4% | measured |
| APEX-Agents | 36.4% | estimated ± 9.4 pp, low confidence |
| APEX-Agents-AA | 27.0% | estimated ± 9.3 pp, medium confidence |
| BioMysteryBench (human-difficult) | 46.7% | estimated ± 6.8 pp, low confidence |
| DeepSearchQA | 92.1% | estimated ± 4.5 pp, high confidence |
| SWE-bench Verified | 84.5% | estimated ± 3.6 pp, high confidence |
| AutomationBench | 30.2% | estimated ± 7.1 pp, low confidence |
| BFCL v4 | 76.2% | estimated ± 2.7 pp, medium confidence |
| CyberGym | 81.9% | estimated ± 3.1 pp, low confidence |
| JobBench | 46.1% | estimated ± 8.7 pp, medium confidence |
| SWE-bench Pro | 64.7% | measured |
| WideResearch | 79.6% | estimated ± 3.3 pp, high confidence |
| AA Agentic Index | 42.1% | measured |
| MMLU-Pro | 87.2% | estimated ± 4.8 pp, high confidence |
| FailureSensorIQ | 67.3% | estimated ± 1.7 pp, high confidence |
| GPQA | 93.1% | estimated ± 0.9 pp, high confidence |
| MMLU | 90.5% | estimated ± 0.9 pp, medium confidence |
| Vals GPQA Diamond | 92.9% | measured |
| Vals LiveCodeBench | 87.4% | measured |
| Vals MMLU-Pro | 89.2% | measured |
| MMMLU | 90.3% | estimated ± 0.7 pp, low confidence |
| MMMU | 86.7% | estimated ± 2.7 pp, medium confidence |
| Chess Puzzles | 35.0% | measured |
| AA MMLU-Pro | 89.0% | estimated ± 0.7 pp, medium confidence |
| Gert Labs | 47.2% | measured |
| AA Harvey LAB v1.0 | 92.6% | estimated ± 1.2 pp, medium confidence |
| AA Tau3 Banking | 42.1% | measured |
| AA Terminal-Bench 2.1 | 81.6% | measured |
| ApprenticeBench | 6.2% | estimated ± 10.4 pp, low confidence |
| ARC-AGI-1 | 85.7% | measured |
| ARC-AGI-2 | 52.6% | measured |
| BabaIsBench | 75.0% | measured |
| CursorBench 3.1 | 56.3% | estimated ± 3.4 pp, medium confidence |
| CursorBench 3.2 | 66.7% | measured |
| EBR-bench | 8.2% | estimated ± 9.3 pp, medium confidence |
| Terminal-Bench 3.0 | 15.7% | measured |
| FrontierCode 1.1 Main | 49.9% | estimated ± 1.6 pp, medium confidence |
| FrontierMath v2 (Tier 4) | 24.4% | measured |
| FrontierMath v2 (Tiers 1-3) | 57.2% | measured |
| FrontierSWE v2 | 23.7% | estimated ± 8.3 pp, medium confidence |
| Furniture Assembly | 22.5% | measured |
| MaxxBench | 49.1% | estimated ± 6.6 pp, low confidence |
| MazeBench | 0.5% | estimated ± 2.7 pp, high confidence |
| MirrorCode | 32.2% | measured |
| Mystery Game Puzzles | 26.0% | estimated ± 10.3 pp, low confidence |
| NYT Connections | 79.9% | measured |
| OfficeQA Pro | 59.5% | estimated ± 6.8 pp, medium confidence |
| OSWorld-Verified | 78.8% | estimated ± 8.4 pp, low confidence |
| ParseBench-1 | 65.8% | estimated ± 6.2 pp, medium confidence |
| Terminal-Bench 4.1 | 12.4% | measured |
| Terminal-Bench Hard | 55.9% | estimated ± 2.8 pp, medium confidence |
| Vals SWE-bench | 86.6% | measured |
| Terminal-Bench 2.1 (Vals AI) | 67.8% | measured |
| VulcanBench v3 | 89.9% | measured |
| CWE-Bench-AA | 50.7% | estimated ± 8.1 pp, medium confidence |
| AA Cyber Index | 25.5% | estimated ± 9.2 pp, low confidence |
| DeepsecBench-AA | 16.1% | estimated ± 8.2 pp, medium confidence |
| HealthBench Professional (length-adjusted, Artificial Analysis) | 57.7% | estimated ± 4.3 pp, medium confidence |
| HealthBench Professional (raw, Artificial Analysis) | 62.0% | estimated ± 0.8 pp, low confidence |
| AA ITBench | 46.3% | estimated ± 4.0 pp, high confidence |
| Bug Hunt Bench | 7.0% | estimated ± 6.3 pp, low confidence |
| Chartography (tools) | 83.6% | estimated ± 4.1 pp, low confidence |
| CursorBench 4.0 | 36.6% | estimated ± 2.0 pp, high confidence |
| CWE-bench v1 | 49.0% | estimated ± 5.7 pp, medium confidence |
| DeepSWE | 53.0% | measured |
| DRACO | 82.4% | estimated ± 2.0 pp, medium confidence |
| FrontierCode 2.0 | 25.4% | estimated ± 5.4 pp, low confidence |
| FrontierPhysics | 0.0% | estimated ± 4.4 pp, high confidence |
| HLE w/o tools | 47.8% | estimated ± 4.9 pp, high confidence |
| OSWorld 2.1 | 57.6% | estimated ± 3.6 pp, high confidence |
| OSWorld 2.1 Offline | 42.1% | estimated ± 1.9 pp, low confidence |
| PostTrainBench v1.1 | 23.5% | measured |
| ProgramBench | 0.0% | estimated ± 4.8 pp, low confidence |
| SpireClimb | 5.8% | estimated ± 7.2 pp, medium confidence |
| StealthMark | 45.2% | estimated ± 6.5 pp, medium confidence |
| SWE Multilingual | 78.0% | measured |
| Terminal-Bench 4.0 | 19.9% | estimated ± 4.5 pp, high confidence |
| Terminal-Bench-Science 0.1 | 0.0% | estimated ± 10.7 pp, low confidence |
| Toolathlon-Verified | 69.5% | estimated ± 2.3 pp, medium confidence |
| VulcanBench Frontier v4 | 88.4% | estimated ± 2.3 pp, medium confidence |
| AA AutomationBench | 52.7% | estimated ± 9.9 pp, low confidence |
| GDP.pdf | 19.1% | estimated ± 4.2 pp, high confidence |
| AA Terminal-Bench 4.0 | 12.0% | estimated ± 6.2 pp, medium confidence |
| ARC-AGI-3 | 0.3% | measured |
| ARC-AGI-3 Provider Adapter | 1.6% | estimated ± 4.1 pp, medium confidence |
| cursorBench41 | 44.8% | estimated ± 3.7 pp, high confidence |
| CWE-bench | 45.2% | estimated ± 6.2 pp, medium confidence |
| MKA | 35.0% | estimated ± 4.6 pp, high confidence |
| SkillsBench official leaderboard (BenchFlow) | 54.3% | estimated ± 6.1 pp, medium confidence |
| BinaryAudit | 35.8% | estimated ± 12.1 pp, low confidence |
| OTelBench | 22.5% | estimated ± 6.6 pp, medium confidence |
| Vibe Code Bench | 61.2% | estimated ± 10.0 pp, low confidence |
| FACTS Multimodal | 45.1% | estimated ± 4.0 pp, high confidence |
| Hermes Bench | 67.9% | estimated ± 2.9 pp, high confidence |
| Hermes Index | 30.9% | estimated ± 5.7 pp, low confidence |
| ParseBench-2 | 72.4% | estimated ± 6.2 pp, medium confidence |
| CharXiv | 84.3% | estimated ± 6.3 pp, low confidence |
| CharXiv w/o tools | 83.5% | estimated ± 2.5 pp, high confidence |
| ExploitBench | 32.4% | estimated ± 6.5 pp, low confidence |
| ExploitGym | 3.1% | estimated ± 8.3 pp, low confidence |
| AI2D_TEST | 94.1% | estimated ± 2.0 pp, low confidence |
| C-Eval | 97.2% | estimated ± 1.0 pp, low confidence |
| CountBench | 98.4% | estimated ± 2.9 pp, low confidence |
| DynaMath | 85.5% | estimated ± 3.0 pp, low confidence |
| ERQA | 68.7% | estimated ± 3.1 pp, high confidence |
| HMMT Feb 2026 | 92.7% | estimated ± 1.5 pp, medium confidence |
| HMMT Nov 2025 | 94.9% | estimated ± 1.5 pp, medium confidence |
| INCLUDE | 85.9% | estimated ± 2.0 pp, medium confidence |
| MathVision | 90.9% | estimated ± 3.1 pp, high confidence |
| MCP Atlas | 77.9% | estimated ± 6.0 pp, low confidence |
| MMAnswerBench | 88.4% | estimated ± 2.4 pp, medium confidence |
| MMLU-ProX | 85.9% | estimated ± 1.3 pp, high confidence |
| MMLU-Redux | 97.4% | estimated ± 0.8 pp, low confidence |
| MMMU-Pro | 80.9% | estimated ± 1.5 pp, high confidence |
| NL2Repo | 50.6% | estimated ± 4.7 pp, high confidence |
| PostTrainBench v1 | 25.6% | estimated ± 1.3 pp, medium confidence |
| QwenClawBench | 93.2% | estimated ± 6.0 pp, low confidence |
| RealWorldQA | 86.4% | estimated ± 2.9 pp, medium confidence |
| ScreenSpot Pro | 77.7% | estimated ± 10.8 pp, low confidence |
| SuperGPQA | 100.0% | estimated ± 9.3 pp, low confidence |
| Terminal-Bench 2.0 | 71.6% | estimated ± 5.8 pp, medium confidence |
| Toolathlon | 55.9% | estimated ± 6.3 pp, medium confidence |
| V* | 100.0% | estimated ± 7.5 pp, low confidence |
| Video-MME (w/o subtitle) | 87.0% | estimated ± 1.7 pp, low confidence |
| Video-MME (with subtitle) | 88.5% | estimated ± 3.7 pp, medium confidence |
| AA LiveCodeBench | 95.6% | estimated ± 2.6 pp, medium confidence |
| HealthBench Hard | 31.4% | estimated ± 8.6 pp, low confidence |
| LiveCodeBench Pro | 87.3% | estimated ± 6.0 pp, low confidence |
| MedXpertQA (MM) | 77.1% | estimated ± 6.5 pp, low confidence |
| React Native Evals | 82.3% | estimated ± 3.2 pp, high confidence |
| ResearchClawBench | 20.4% | estimated ± 1.9 pp, high confidence |
| PostTrainBench | 22.6% | measured |
| DAYJOB | 1.4% | estimated ± 2.7 pp, medium confidence |
| DAYJOB Healthcare | 0.0% | estimated ± 4.0 pp, medium confidence |
| Finance Agent v2 | 55.6% | estimated ± 2.4 pp, medium confidence |
| BioMysteryBench (human-solvable) | 87.8% | estimated ± 0.6 pp, medium confidence |
| FrontierCode 1.1 Extended | 59.4% | estimated ± 1.6 pp, medium confidence |
| HealthBench (raw) | 51.2% | estimated ± 6.0 pp, medium confidence |
| HealthBench (length-adjusted) | 53.8% | estimated ± 0.9 pp, medium confidence |
| HealthBench Professional | 56.8% | estimated ± 3.6 pp, high confidence |
| HealthBench Professional (raw) | 61.6% | estimated ± 4.5 pp, high confidence |
| HLE-Verified | 53.5% | estimated ± 2.0 pp, medium confidence |
| LABBench2 | 82.7% | estimated ± 1.8 pp, low confidence |
| ScarfBench | 61.7% | estimated ± 3.8 pp, low confidence |
| BlenderWorld (text-only) | 70.6% | estimated ± 4.0 pp, medium confidence |
| Cookie-Bench | 81.5% | estimated ± 2.5 pp, medium confidence |
| STEMBench | 2.6% | estimated ± 0.2 pp, medium confidence |
| SimpleQA | 35.2% | estimated ± 7.8 pp, low confidence |
| Agents' Last Exam | 30.9% | estimated ± 9.7 pp, medium confidence |
| BabyVision w/ Python | 91.1% | estimated ± 4.0 pp, low confidence |
| OpenHarmony Bench v1.0 | 58.9% | estimated ± 2.4 pp, high confidence |
| SEC-Bench Pro | 33.0% | estimated ± 10.7 pp, low confidence |
| MRCRv2 | 85.7% | estimated ± 11.3 pp, low confidence |
| MMLongBench-Doc | 61.4% | estimated ± 2.6 pp, low confidence |
| BioBasicsBench | 34.1% | estimated ± 2.8 pp, medium confidence |
| LifeVal | 49.0% | estimated ± 5.0 pp, low confidence |
| MMMU-Pro w/ Python | 83.1% | estimated ± 2.0 pp, high confidence |
| ExploitBench Internal Port | 0.2% | estimated ± 1.4 pp, low confidence |
| SWE-Atlas Codebase QnA | 49.8% | estimated ± 4.8 pp, high confidence |