xAI · model
Grok 4.6 benchmark scores
As of 2026-10-07, Grok 4.6 (xAI) has measured scores on 31 benchmarks and estimated scores on 63 more.
| Benchmark | Score | Source |
|---|---|---|
| AA-GPQA Diamond | 94.9% | measured |
| AA-HLE | 42.9% | measured |
| AA-SciCode | 56.5% | measured |
| Artificial Analysis Intelligence Index | 44.3% | measured |
| CritPt | 17.1% | measured |
| GDPval-AA | 56.1% | measured |
| GPQA Diamond | 90.0% | estimated ± 1.4 pp, high confidence |
| HLE | 60.0% | estimated ± 5.5 pp, low confidence |
| AA-LCR | 80.3% | measured |
| LiveCodeBench v6 | 93.6% | estimated ± 3.1 pp, medium confidence |
| AA-Omniscience Accuracy | 48.2% | measured |
| SciCode | 53.1% | estimated ± 2.9 pp, low confidence |
| Terminal-Bench 2.1 | 85.8% | estimated ± 6.6 pp, low confidence |
| BrowseComp | 84.5% | estimated ± 2.8 pp, high confidence |
| HLE w/ tools | 60.4% | estimated ± 6.1 pp, medium confidence |
| LongBench v2 | 61.5% | estimated ± 5.9 pp, low confidence |
| τ²-bench results | 96.0% | estimated ± 2.4 pp, medium confidence |
| AA Coding Index | 76.8% | measured |
| APEX-Agents | 57.5% | measured |
| APEX-Agents-AA | 36.9% | estimated ± 3.5 pp, low confidence |
| BioMysteryBench (human-difficult) | 44.8% | estimated ± 5.9 pp, low confidence |
| DeepSearchQA | 93.8% | estimated ± 4.5 pp, high confidence |
| SWE-bench Verified | 90.5% | estimated ± 4.4 pp, high confidence |
| AutomationBench | 35.0% | estimated ± 7.0 pp, low confidence |
| BFCL v4 | 77.9% | estimated ± 2.6 pp, medium confidence |
| CyberGym | 77.1% | estimated ± 5.0 pp, medium confidence |
| JobBench | 59.5% | estimated ± 8.6 pp, medium confidence |
| SWE-bench Pro | 69.4% | estimated ± 4.3 pp, high confidence |
| WideResearch | 100.0% | estimated ± 9.1 pp, low confidence |
| AA Agentic Index | 53.4% | measured |
| MMLU-Pro | 85.7% | estimated ± 3.5 pp, high confidence |
| GPQA | 93.7% | estimated ± 0.8 pp, high confidence |
| Gert Labs | 68.0% | estimated ± 3.0 pp, medium confidence |
| AA-AnalystAgent | 41.8% | estimated ± 3.6 pp, medium confidence |
| AA EnterpriseOps-Gym | 48.3% | measured |
| AA Harvey LAB | 93.6% | estimated ± 1.2 pp, medium confidence |
| ApprenticeBench | 13.0% | measured |
| ARC-AGI-1 | 87.0% | measured |
| ARC-AGI-2 | 67.1% | measured |
| CursorBench 3.1 | 60.0% | estimated ± 3.4 pp, medium confidence |
| CursorBench 3.2 | 70.8% | measured |
| Terminal-Bench 3.0 | 26.5% | measured |
| FrontierCode 1.1 Main | 54.7% | estimated ± 1.6 pp, low confidence |
| FrontierSWE v2 | 25.3% | measured |
| OSWorld-Verified | 76.8% | estimated ± 1.9 pp, medium confidence |
| Vals GPQA Diamond | 94.7% | measured |
| Vals LiveCodeBench | 88.2% | measured |
| Vals MMLU-Pro | 89.4% | measured |
| Vals SWE-bench | 95.6% | measured |
| Terminal-Bench 2.1 (Vals AI) | 78.3% | measured |
| VulcanBench v3 | 87.0% | measured |
| AA AutomationBench | 66.7% | measured |
| GDP.pdf | 16.7% | estimated ± 3.7 pp, low confidence |
| AA ITBench | 33.3% | estimated ± 5.7 pp, low confidence |
| AA Tau3 Banking | 50.7% | measured |
| AA Terminal-Bench 2.1 | 87.6% | estimated ± 4.4 pp, high confidence |
| AA Terminal-Bench 4.0 | 34.4% | estimated ± 14.0 pp, low confidence |
| CursorBench 4.0 | 41.4% | measured |
| CWE-bench v1 | 57.0% | measured |
| DeepSWE | 65.9% | measured |
| HLE w/o tools | 51.4% | estimated ± 5.0 pp, medium confidence |
| OSWorld 2.0 | 59.6% | estimated ± 13.3 pp, low confidence |
| PostTrainBench v1.1 | 33.8% | estimated ± 2.7 pp, high confidence |
| ProgramBench | 89.4% | estimated ± 6.0 pp, low confidence |
| Toolathlon-Verified | 75.9% | estimated ± 2.3 pp, medium confidence |
| Vibe Code Bench | 75.0% | estimated ± 10.3 pp, low confidence |
| ExploitGym | 15.5% | estimated ± 10.5 pp, low confidence |
| C-Eval | 96.1% | estimated ± 1.0 pp, low confidence |
| Claw-Eval | 85.3% | estimated ± 13.6 pp, low confidence |
| MCP Atlas | 100.0% | estimated ± 0.7 pp, low confidence |
| MMLU-Redux | 97.3% | estimated ± 1.1 pp, low confidence |
| NL2Repo | 59.4% | estimated ± 4.7 pp, high confidence |
| QwenClawBench | 100.0% | estimated ± 6.0 pp, low confidence |
| SuperGPQA | 85.0% | estimated ± 7.5 pp, low confidence |
| Terminal-Bench 2.0 | 78.8% | estimated ± 5.9 pp, low confidence |
| Toolathlon | 54.3% | estimated ± 1.7 pp, medium confidence |
| HealthBench Hard | 34.3% | estimated ± 7.7 pp, low confidence |
| MMLU-Pro (Arcee) | 100.0% | estimated ± 4.0 pp, low confidence |
| React Native Evals | 100.0% | estimated ± 2.4 pp, low confidence |
| ResearchClawBench | 21.2% | estimated ± 2.0 pp, medium confidence |
| SWE-Rebench | 58.1% | estimated ± 8.9 pp, low confidence |
| ARC-AGI-3 | 2.1% | measured |
| BioMysteryBench (human-solvable) | 88.5% | estimated ± 0.6 pp, medium confidence |
| DRACO | 84.6% | estimated ± 3.0 pp, medium confidence |
| FrontierCode 1.1 Extended | 61.3% | measured |
| HealthBench (raw) | 51.7% | estimated ± 6.4 pp, low confidence |
| HealthBench (length-adjusted) | 53.8% | estimated ± 0.9 pp, medium confidence |
| HealthBench Professional | 59.9% | estimated ± 1.3 pp, medium confidence |
| HealthBench Professional (raw) | 61.9% | estimated ± 5.0 pp, medium confidence |
| HLE-Verified | 35.3% | estimated ± 0.7 pp, low confidence |
| LABBench2 | 65.7% | estimated ± 1.3 pp, low confidence |
| LiveCodeBench | 100.0% | estimated ± 8.9 pp, low confidence |
| OpenHarmony Bench | 58.8% | estimated ± 2.4 pp, high confidence |
| MMLU | 94.7% | estimated ± 1.1 pp, low confidence |