Introducing Grok 4.5 16 Jul 2026
Grok 4.5
Grok 4.5 ranks first on 2 of the 123 benchmarks it has scores on.
142
Benchmarks with a score
2
Ranked first, out of five or more models
5
In the top three
xAI Jul 2026
Grok 4.5
Closed weights Flagship
154.4
#44 of 271
Epoch Capabilities Index
- DeepSWE 53.8% #19 /28
- Terminal-Bench 4.0 12.4% #22 /23
- τ³-bench Banking 47.9% #3 /25
Where it ranks
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.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LMArena (Elo) | 1,465 | #43 of 333 | Official leaderboard | |
| BullshitBench | 60% | #18 of 107 | Official leaderboard | |
| SimpleQA Verified | 53.8% | #13 of 56 | Official leaderboard | |
| Judgemark v4 | 77.1 | #10 of 40 | Official leaderboard | |
| Global MMLU Lite | 93.3% | #13 of 43 | Official leaderboard | |
| BullshitBench V2 | 55% | #38 of 122 | Official leaderboard | |
| Extended NYT Connections | 79.9% | #39 of 97 | Official leaderboard | |
| Creative Writing v3 | 70.5 | #51 of 120 | Official leaderboard | |
| Document Arena | 1,464 | #18 of 37 | Official leaderboard | |
| FACTS Parametric | 46.8% | #22 of 45 | Official leaderboard | |
| FACTS Benchmark Suite | 40.6% | #25 of 40 | Official leaderboard | |
| FACTS Search V2 | 79.4% | #14 of 21 | Official leaderboard | |
| FACTS Parametric V2 | 42.5% | #17 of 20 | Official leaderboard | |
| xAI Factuality (Hallucination) eval | 0.98% | #2 of 2 | Model card |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| DTBench | 94.2% | #11 of 178 | Official leaderboard | |
| GPQA Diamond | 93.4% | #17 of 219 | Epoch AI | |
| OTIS Mock AIME 2024-2025 | 97.8% | #33 of 202 | Epoch AI | |
| SimpleBench | 70% | #20 of 100 | Official leaderboard | |
| BixBench | 93.8% | #1 of 4 | Model card | |
| CritPt | 15.4% | #43 of 160 | Epoch AI | |
| Surface Evolver Bench | 74.4% | #8 of 26 | Epoch AI | |
| xbench-ScienceQA | 72.4% | #9 of 29 | Official leaderboard | |
| ProofBench | 30% | #23 of 59 | Official leaderboard | |
| ARC-AGI | 87.2% | #40 of 92 | Official leaderboard | |
| ARC-AGI-2 | 52.6% | #42 of 95 | Official leaderboard | |
| FrontierMath Tiers 1-3 v2 | 57.2% | #47 of 89 | Epoch AI | |
| FrontierMath Tier 4 v2 | 24.4% | #45 of 71 | Epoch AI | |
| ProofBench v1.1 | 31% | #36 of 51 | Official leaderboard | |
| ForecastBench | 65.9 | #82 of 86 | Official leaderboard |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SWE-Marathon | 29% | #1 of 12 | Official leaderboard | |
| SWE Atlas | 84% | #1 of 11 | Model card | |
| SWE-rebench | 63.8% | #2 of 13 | Official leaderboard | |
| GBAEval | 65.4% | #4 of 21 | Official leaderboard | |
| SWE-Bench Pro | 64.7% | #15 of 77 | Model card | |
| SWE-bench Multilingual | 78% | #18 of 70 | Model card | |
| ALE-Bench | 1,309 | #32 of 121 | Epoch AI | |
| WebDev Arena | 1,555 | #31 of 105 | Epoch AI | |
| Code Arena | 1,553 | #40 of 124 | Official leaderboard | |
| Vibe Code Bench | 69% | #47 of 115 | Official leaderboard | |
| Code Migration | 36.6% | #35 of 78 | Official leaderboard | |
| Image-to-WebDev Arena | 1,562 | #25 of 55 | Official leaderboard | |
| FrontierCode | 42.4% | #21 of 46 | Epoch AI | |
| WeirdML | 46.4% | #55 of 112 | Official leaderboard | |
| ProgramBench | 57.2% | #11 of 20 | Model card | |
| Senior SWE-Bench | 23.2% | #14 of 23 | Official leaderboard | |
| DeepSWE | 53.8% | #19 of 28 | Official leaderboard | |
| MirrorCode | 32.2% | #21 of 28 | Epoch AI | |
| IOI v2 (Vals) | 40.6% | #34 of 45 | Official leaderboard | |
| PostTrainBench | 23.4% | #9 of 11 | Epoch AI | |
| PostTrainBench 1.1 | 23.4% | #9 of 11 | Official leaderboard | |
| PostTrainBench 1.2 | 22.6% | #15 of 17 | Official leaderboard | |
| APEX-SWE | 51.2% | #2 of 2 | Model card | |
| RelBench | 38.4% | #1 of 1 | Model card |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| τ³-bench Banking | 47.9% | #3 of 25 | Official leaderboard | |
| Long-Horizon-Terminal-Bench (LHTB) | 50.5% | #4 of 25 | Official leaderboard | |
| Game Arena Bargaining | 223 | #10 of 36 | Official leaderboard | |
| Terminal-Bench 2.1 | 79.3% | #4 of 13 | Official leaderboard | |
| Game Arena Lines of Action | 585 | #12 of 36 | Official leaderboard | |
| Game Arena Ultimate Tic-Tac-Toe | 381 | #12 of 36 | Official leaderboard | |
| LMArena Search Arena | 1,206 | #15 of 35 | Official leaderboard | |
| PinchBench 2.0 | 75.2% | #25 of 57 | Official leaderboard | |
| Game Arena Word Association | 260 | #17 of 36 | Official leaderboard | |
| Game Arena Four in a Row | 367 | #18 of 38 | Official leaderboard | |
| Game Arena Coin Game | 367 | #17 of 35 | Official leaderboard | |
| Game Arena Go | 910 | #18 of 36 | Official leaderboard | |
| Chess Puzzles | 35% | #44 of 83 | Epoch AI | |
| Game Arena Dots and Boxes | 875 | #21 of 37 | Official leaderboard | |
| Game Arena Poker | -0.85 | #18 of 31 | Official leaderboard | |
| Kaggle Game Arena | 258 | #24 of 39 | Official leaderboard | |
| Game Arena Chess | 643 | #26 of 39 | Official leaderboard | |
| Game Arena Nine Men's Morris | 527 | #24 of 36 | Official leaderboard | |
| Game Arena Dark Hex | 173 | #26 of 38 | Official leaderboard | |
| Game Arena Clobber | 152 | #26 of 36 | Official leaderboard | |
| Game Arena Reversi | 239 | #26 of 36 | Official leaderboard | |
| Terminal-Bench 3.0 | 15.7% | #8 of 11 | Official leaderboard | |
| CEO-Bench | $15,909 | #15 of 19 | Official leaderboard | |
| Game Arena Word Art | 178 | #29 of 35 | Official leaderboard | |
| ARC-AGI-3 | 0.32% | #20 of 24 | Official leaderboard | |
| Time Horizon Index: KSP | 6.33% | #11 of 13 | Official leaderboard | |
| Game Arena Checkers | 83 | #36 of 38 | Official leaderboard | |
| DeepSearchQA | 38.4% | #21 of 22 | Model card | |
| Terminal-Bench 4.0 | 12.4% | #22 of 23 | Official leaderboard | |
| Vals Web Search Index | 38.8% | #4 of 4 | Official leaderboard |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CorpFin v2 | 67.4% | #13 of 119 | Official leaderboard | |
| Harvey LAB (Legal Agent Benchmark) | 12.9% | #11 of 80 | Official leaderboard | |
| LegalBench | 86% | #19 of 138 | Official leaderboard | |
| MedScribe | 86.9% | #19 of 98 | Official leaderboard | |
| APEX-Agents | 56.2% | #19 of 56 | Epoch AI | |
| ProfBench | 51% | #27 of 72 | Official leaderboard | |
| RadLE 2.0 | 524.0 | #5 of 12 | Official leaderboard | |
| PolicyBench | 84.7% | #22 of 50 | Official leaderboard | |
| Legal Research Bench | 38% | #39 of 88 | Official leaderboard | |
| MedCode | 43.3% | #44 of 96 | Official leaderboard | |
| Tax Agent Bench | 24.9% | #42 of 79 | Official leaderboard | |
| GDPval-AA v2 | 1,535 | #14 of 26 | Model card | |
| TaxEval v2 | 71.7% | #72 of 129 | Official leaderboard | |
| EEBench | 50.9% | #16 of 26 | Official leaderboard | |
| Vending-Bench 2 | $3,887 | #41 of 66 | Official leaderboard | |
| Vals Finance Agent v2 | 48.4% | #50 of 80 | Official leaderboard | |
| MortgageTax | 61.8% | #55 of 86 | Official leaderboard | |
| Excel Modeling Benchmark | 52.9% | #52 of 76 | Official leaderboard | |
| $OneMillion-Bench | 15.8% | #12 of 13 | Official leaderboard | |
| Parametric CAD Bench | 88.1% | #3 of 3 | Model card | |
| 3DCodeBench | 49.8% | #2 of 2 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vision Arena | 1,279 | #24 of 136 | Official leaderboard | |
| ZeroBench | 15% | #21 of 66 | Official leaderboard | |
| ClockBench | 21.7% | #18 of 37 | Official leaderboard | |
| VoxelBench | 1,739 | #28 of 56 | Official leaderboard | |
| Blueprint-Bench 2 | 27.3% | #22 of 35 | Official leaderboard | |
| GDP.pdf | 14% | #32 of 39 | Epoch AI | |
| ArtifactArena | 957 | #18 of 21 | Official leaderboard | |
| Furniture Assembly Benchmark | 22.5% | #31 of 35 | Epoch AI |
Safety and security
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Virology Capabilities Test (VCT) | 65.5% | #2 of 28 | Model card | |
| Child safety (xAI) | 0% | #1 of 3 | Model card | |
| FORTRESS-RN | 97.9% | #1 of 3 | Model card | |
| HackerBench | 7.80% | #1 of 3 | Model card | |
| ProtocolQA Open-Ended (xAI) | 87% | #1 of 3 | Model card | |
| Self-harm refusals (xAI) | 0.50% | #1 of 3 | Model card | |
| Human Pathogen Capabilities Test v2 (HPCT v2) | 61.2% | #26 of 71 | Official leaderboard | |
| CyberGym | 79% | #13 of 33 | Official leaderboard | |
| CyberBench v1.1 (Vals) | 71% | #18 of 43 | Official leaderboard | |
| Dictatorship Eval | 66% | #11 of 20 | Official leaderboard | |
| DeepsecBench | 16.4 | #18 of 28 | Official leaderboard | |
| General refusals (xAI) | 1.10% | #5 of 7 | Model card | |
| PCT (Polarized Contrastive Pairs) | 50.2% | #34 of 46 | Official leaderboard | |
| SRE-Bench | 0.76% | #25 of 33 | Official leaderboard | |
| MASK | 0.67% | #20 of 21 | Model card | |
| Autointent-Bio | 97.9% | #3 of 3 | Model card | |
| Autointent-Chem | 96.7% | #3 of 3 | Model card | |
| LAB-Bench | 71.1% | #3 of 3 | Model card | |
| Standard jailbreaks (xAI) | 0.73% | #3 of 3 | Model card | |
| xAI sycophancy eval | 0.01% | #3 of 3 | Model card | |
| Epistemic Bias (xAI) | 20.4% | #1 of 1 | Model card | |
| FalseClaimBench | 84% | #1 of 1 | Model card |
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Conceptual Reasoning Index | 60.3 | #32 of 149 | Official leaderboard | |
| ACCoRD | 71.6 | #36 of 150 | Official leaderboard | |
| LMCA | 45.2% | #38 of 139 | Epoch AI | |
| CAIS Vision Capabilities Index | 56.2 | #21 of 53 | Official leaderboard | |
| Vals Multimodal Index | 63.4% | #15 of 33 | Official leaderboard | |
| CAIS Text Capabilities Index | 32.3 | #32 of 66 | Official leaderboard | |
| CAIS Risk Index | 53.4 | #36 of 65 | Official leaderboard | |
| SAGE (Vals) | 35% | #68 of 81 | Official leaderboard |
xAI's models over time
xAI's models in color, against every other model on the Epoch Capabilities Index, by release date.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it. 32 of them aren't in the launch post.
Model Card: Grok 4.5 14 Jul 2026
Capabilities
- DeepSWE 62
- APEX-SWE 51.2
- SWE-Bench Pro 64.7
- SWE-bench Multilingual 78
- SWE-Marathon 29
- FrontierSWE v1 78
- ProgramBench 57.2
- Terminal-Bench 2.1 83.3
- SWE Atlas 84
- EEBench
- 3DCodeBench 49.8
- Parametric CAD Bench 88.1
- GDPval-AA v2 1535
- τ³-bench Banking 32.6
- DeepSearchQA 38.4
Hallucination
Safety
- General refusals (xAI) 1.1
- Child safety (xAI) 0
- Autointent-Bio 97.9
- Autointent-Chem 96.7
- FORTRESS-RN 97.9
- Self-harm refusals (xAI) 0.5
Robustness
Bias
- Epistemic Bias (xAI) 20.4
Alignment
- FalseClaimBench 84
- MASK 0.67
- xAI sycophancy eval 0.01
Bio and chem risk
- Virology Capabilities Test (VCT) 65.5
- WMDP
- LAB-Bench 71.1
- ProtocolQA Open-Ended (xAI) 87
- BixBench 93.8
Cyber risk
- CyberGym 80.4
- HackerBench 7.8