Grok 4 ranks first on 6 of the 159 benchmarks with five or more models scored.
Closed weights
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| IFEval | 94.9% | #2 of 60 | Official leaderboard | |
| Fiction.LiveBench | 94.4% | #3 of 57 | Fiction.live leaderboard, via Epoch AI | |
| MMLU-Pro | 87% | #18 of 212 | Official leaderboard | |
| SuperGPQA | 63% | #6 of 59 | Official leaderboard | |
| LLM Confabulation Leaderboard | 12.4% | #9 of 58 | Official leaderboard | |
| SimpleQA Verified | 51.9% | #17 of 56 | Official leaderboard | |
| FACTS Parametric | 58.6% | #14 of 45 | Official leaderboard | |
| FACTS Search | 75.3% | #10 of 31 | Official leaderboard | |
| LMArena (Elo) | 1,411 | #120 of 335 | Official leaderboard | |
| HUMAINE | 30 | #18 of 50 | Official leaderboard | |
| WritingBench | 74.7 | #18 of 45 | Official leaderboard | |
| TableBench | 57.8% | #4 of 10 | Official leaderboard | |
| Humanity's Last Exam | 38.6% | #66 of 154 | Lab model card | |
| IndicGenBench | 52.0 | #7 of 13 | Official leaderboard | |
| Global MMLU Lite | 88.8% | #24 of 43 | Official leaderboard | |
| SkateBench v2 | 61.3% | #12 of 21 | Official leaderboard | |
| EQ-Bench 3 | 1,154 | #47 of 78 | Official leaderboard | |
| Vectara Hallucination Leaderboard (HHEM) | 4.80% | #92 of 148 | Official leaderboard | |
| ConStory-Bench | 0.67 | #9 of 14 | Official leaderboard | |
| DiscoX | 43.8 | #5 of 7 | Official leaderboard | |
| Phare | 62.1% | #46 of 62 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 45.7 | #92 of 119 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MATH-500 | 96.2% | #2 of 38 | Official leaderboard | |
| AIME 2025 | 98.8% | #12 of 143 | Lab model card | |
| PHYBench | 42.3% | #3 of 28 | Official leaderboard | |
| GPQA Main | 72.7% | #7 of 60 | Official leaderboard | |
| FutureX | 27.9 | #1 of 7 | Official leaderboard | |
| HiPhO | 64.4% | #7 of 36 | Official leaderboard | |
| ForecastBench | 0.13 | #18 of 82 | Official leaderboard | |
| SciArena | 1,045 | #8 of 34 | Official leaderboard | |
| HMMT 2025 | 93.9% | #15 of 54 | Lab model card | |
| BRUMO 2025 | 95% | #13 of 42 | Official leaderboard | |
| GPQA Diamond | 87% | #72 of 229 | Epoch AI run | |
| IneqMath | 8.00% | #12 of 34 | Official leaderboard | |
| CMIMC 2025 | 85% | #12 of 33 | Official leaderboard | |
| SimpleBench | 60.5% | #42 of 106 | Official leaderboard | |
| SMT 2025 | 85.8% | #17 of 40 | Official leaderboard | |
| FrontierScience-Research | 15.9% | #3 of 7 | Official leaderboard | |
| OTIS Mock AIME 2024-2025 | 84% | #91 of 212 | Epoch AI run | |
| FrontierMath | 19.7% | #31 of 70 | Epoch AI run | |
| SnorkelSequences | 63.2% | #12 of 25 | Official leaderboard | |
| SnorkelGraph | 61% | #12 of 24 | Official leaderboard | |
| ARC-AGI-2 | 16% | #56 of 98 | Official leaderboard | |
| FrontierScience-Olympiad | 66.2% | #4 of 7 | Official leaderboard | |
| ARC-AGI | 66.7% | #55 of 95 | Official leaderboard | |
| IMO-ProofBench | 18.6% | #12 of 20 | Official leaderboard | |
| ATLAS (OpenCompass) | 34.1% | #3 of 5 | Official leaderboard | |
| MathArena (Apex, IMO, USAMO) | 2.08% | #29 of 47 | Official leaderboard | |
| MathArena Apex Shortlist | 57.5% | #24 of 38 | Official leaderboard | |
| FrontierMath Tier 4 | 2.08% | #40 of 55 | Epoch AI run | |
| HMMT Nov 2025 | 88.3% | #16 of 22 | Official leaderboard | |
| SGI-Bench | 28.7% | #11 of 15 | Official leaderboard | |
| IMProofBench | 11.1% | #7 of 9 | Official leaderboard | |
| IMO 2025 | 11.9% | #5 of 6 | Official leaderboard | |
| xbench-ScienceQA | 65% | #28 of 29 | Official leaderboard | |
| SpatialBench (LatchBio) | 31.9% | #14 of 14 | Official leaderboard | |
| USAMO 2025 | 37.5% | #2 of 2 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Aider Polyglot | 79.6% | #5 of 46 | Official leaderboard | |
| AutoCodeBench | 50.9% | #4 of 34 | Official leaderboard | |
| METR Time Horizon | 105 min | #6 of 31 | Official leaderboard | |
| IOI (Vals) | 26.2% | #17 of 55 | Official leaderboard | |
| WeirdML | 45.7% | #59 of 112 | Official leaderboard | |
| FrontierCS | 24.0 | #11 of 14 | Official leaderboard | |
| DesignArena 3D Design | 1,034 | #115 of 136 | Official leaderboard | |
| DesignArena SVG | 1,011 | #90 of 103 | Official leaderboard | |
| DesignArena Game Dev | 992 | #125 of 143 | Official leaderboard | |
| DesignArena Website | 1,007 | #132 of 151 | Official leaderboard | |
| DesignArena Code | 1,006 | #126 of 143 | Official leaderboard | |
| DesignArena UI Component | 977 | #123 of 139 | Official leaderboard | |
| OTelBench | 4.35% | #13 of 14 | Official leaderboard | |
| DesignArena Data Visualization | 1,029 | #110 of 118 | Official leaderboard | |
| APEX-SWE | 21% | #62 of 65 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SnorkelWordle | 93% | #2 of 19 | Official leaderboard | |
| BFCL V4 | 63% | #8 of 65 | Official leaderboard | |
| lmgame-Bench | 52.1 | #3 of 23 | Official leaderboard | |
| Game Arena Poker | 16.3 | #5 of 31 | Official leaderboard | |
| MCP-Universe | 33.3% | #5 of 30 | Official leaderboard | |
| C3-Bench | 10.2% | #6 of 30 | Official leaderboard | |
| Elimination Game | 5 | #12 of 56 | Official leaderboard | |
| LisanBench | 4042.2 | #29 of 133 | Official leaderboard | |
| Step Race | 3 | #13 of 59 | Official leaderboard | |
| MCPMark | 31.7% | #7 of 28 | Official leaderboard | |
| CompileBench | 86.7% | #5 of 20 | Official leaderboard | |
| BALROG | 43.6% | #9 of 33 | Official leaderboard | |
| BalatroBench | 9.47 | #6 of 16 | Official leaderboard | |
| Game Arena Chess Openings | 739 | #5 of 11 | Official leaderboard | |
| DeepResearch Bench | 47.3% | #11 of 22 | evals.futuresearch.ai, via Epoch AI | |
| Deep Research Bench (FutureSearch) | 0.47 | #11 of 21 | Official leaderboard | |
| Galileo Agent Leaderboard v2 | 42% | #11 of 21 | Official leaderboard | |
| Meta Agents Research Environments (ARE) / Gaia2 | 15.7% | #7 of 13 | Official leaderboard | |
| Terminal-Bench | 27.2% | #10 of 18 | tbench.ai, via Epoch AI | |
| Game Arena Chess | 756 | #24 of 42 | Official leaderboard | |
| Game Arena Five in a Row | 442 | #12 of 20 | Official leaderboard | |
| Toolathlon | 27.5% | #34 of 50 | Official leaderboard | |
| Game Arena Four in a Row | 265 | #30 of 43 | Official leaderboard | |
| Terminal-Bench 2.0 | 27.2% | #34 of 43 | Official leaderboard | |
| Kaggle Game Arena | 187 | #34 of 42 | Official leaderboard | |
| LMArena Search Arena | 1,123 | #36 of 37 | Official leaderboard | |
| Game Arena Word Association | 0 | #40 of 40 | Official leaderboard | |
| TextQuests | 30.4 | #28 of 28 | Official leaderboard | |
| Game Arena Werewolf | -6 | #8 of 8 | Official leaderboard | |
| Research-Eval | 26.7% | #3 of 3 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Finance Reasoning (Snorkel) | 53.1% | #1 of 21 | Official leaderboard | |
| Vending-Bench | $4,694 | #1 of 9 | Official leaderboard | |
| SnorkelUnderwrite | 83.3% | #4 of 27 | Official leaderboard | |
| CaseLaw v2 | 65.8% | #9 of 53 | Official leaderboard | |
| CorpFin v2 | 66% | #24 of 119 | Official leaderboard | |
| Vals Finance Agent v1 | 40.3% | #18 of 47 | Official leaderboard | |
| Vals Finance Agent | 53.5% | #19 of 49 | Official leaderboard | |
| LegalBench | 83.2% | #57 of 138 | Official leaderboard | |
| FinQA (Snorkel OpenEnv environment) | 74% | #10 of 22 | Official leaderboard | |
| SnorkelFinance | 74% | #10 of 22 | Official leaderboard | |
| ProfBench | 45.1% | #43 of 72 | Official leaderboard | |
| MedScribe | 78.2% | #61 of 97 | Official leaderboard | |
| BizFinBench v2 | 37.2% | #6 of 9 | Official leaderboard | |
| MedCode | 38.1% | #67 of 96 | Official leaderboard | |
| APEX-Agents | 15.2% | #6 of 8 | Official leaderboard | |
| prinzbench | 23.0 | #20 of 26 | Official leaderboard | |
| TaxEval v2 | 65.1% | #106 of 129 | Official leaderboard | |
| Finch (FinWorkBench) | 23.8% | #5 of 6 | Official leaderboard | |
| APEX-1 | 57.1% | #60 of 71 | Official leaderboard | |
| MortgageTax | 44.5% | #79 of 87 | Official leaderboard | |
| GDPval | 24.3% | #5 of 5 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Blueprint-Bench | 0.39 | #4 of 8 | Official leaderboard | |
| ZeroBench | 6.00% | #49 of 76 | Official leaderboard | |
| Vision Arena | 1,184 | #89 of 136 | Official leaderboard | |
| Butter-Bench | 23% | #3 of 4 | Official leaderboard | |
| BabyVision | 16.2% | #23 of 27 | Official leaderboard | |
| Math Kangaroo 2025 (MathArena) | 70% | #19 of 22 | Official leaderboard | |
| GeoBench | 45% | #44 of 48 | Official leaderboard | |
| MeasureBench Real-world | 7.50% | #19 of 19 | Official leaderboard | |
| MeasureBench Synthetic | 6.20% | #19 of 19 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Cybench | 43% | #2 of 16 | Official leaderboard | |
| General refusals (xAI) | 0% | #1 of 8 | Lab model card | |
| Lakera B3 (Backbone Breaker Benchmark) | 79.5% | #5 of 35 | Official leaderboard | |
| Virology Capabilities Test (VCT) | 60% | #5 of 31 | Lab model card | |
| Anthropic political even-handedness eval | 96% | #1 of 5 | Official leaderboard | |
| Refusals + User Jailbreak (xAI) | 0% | #1 of 5 | Lab model card | |
| BioLP-bench | 47% | #2 of 9 | Lab model card | |
| HelpBench | 83% | #2 of 9 | Official leaderboard | |
| BaxBench | 43.6% | #8 of 34 | Official leaderboard | |
| AgentHarm | 14% | #2 of 6 | Lab model card | |
| Molecular Biology Capabilities Test | 51.7% | #33 of 90 | Official leaderboard | |
| AgentDojo | 2.00% | #3 of 8 | Lab model card | |
| SycophancyEval (answer sycophancy) | 7.00% | #2 of 4 | Lab model card | |
| Soft Bias (Internal) | 0.36 | #1 of 2 | Lab model card | |
| Human Pathogen Capabilities Test | 51.4% | #44 of 86 | Official leaderboard | |
| AutoBaxBench | 30.7% | #6 of 11 | Official leaderboard | |
| Human Pathogen Capabilities Test v2 (HPCT v2) | 49.7% | #44 of 71 | Official leaderboard | |
| PCT (Polarized Contrastive Pairs) | 49.5% | #32 of 46 | Official leaderboard | |
| MakeMeSay | 12% | #7 of 10 | Lab model card | |
| MASK | 57% | #15 of 19 | Lab model card | |
| Refusals + System Jailbreak (xAI) | 1.00% | #4 of 5 | Lab model card | |
| AIR-Bench 2024 | 44.4% | #70 of 82 | Official leaderboard | |
| BinaryAudit | 8.77% | #18 of 20 | Official leaderboard | |
| Anthropic Red Team | 95.7% | #68 of 75 | Official leaderboard | |
| HELM Safety | 83.6 | #68 of 75 | Official leaderboard | |
| SimpleSafetyTests | 92.2% | #69 of 75 | Official leaderboard | |
| KORA | 14.8% | #39 of 42 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| HELM Capabilities | 78.5 | #7 of 60 | Official leaderboard | |
| CAIS Text Capabilities Index | 26.2 | #43 of 68 | Official leaderboard | |
| CAIS Risk Index | 59.3 | #50 of 67 | Official leaderboard | |
| CAIS Vision Capabilities Index | 42.5 | #40 of 53 | Official leaderboard | |
| SAGE (Vals) | 25.1% | #79 of 81 | Official leaderboard |
The Self-Reported Index fits one score to the 5 capability results xAI reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 139.8 to 148.8. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| AIME 2025 | 98.8% AIME'25, with Python (no tools 91.7) | 92.5% Leaderboard | +6.3 pts |
| HMMT 2025 | 93.9% HMMT 2025, with Python (no tools 90.0) | 95% Leaderboard | −1.1 pts |
| GPQA Diamond | 87.5% GPQA (Science), no tools | 87% Epoch AI run | +0.5 pts |
| ARC-AGI-2 | 15.9% ARC-AGI-2 | 16% Leaderboard, thinking | Same |
| Cybench | 43% CyBench, unguided success rate 0.43, Grok 4 API, Inspect harness (Table 3) | 43% Leaderboard | Same |
| Vending-Bench | $4,694 Vending-Bench net worth | $4,694 Leaderboard | Same |
xAI's models in color, against every other model on the Epoch Capabilities Index, by release date.
Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 13 of them aren't in the launch post.