xAIGrok
Grok 4
xAI · Grok. Released Jul 9, 2025.
Grok 4 is a model from xAI in the Grok family, released Jul 9, 2025. evals.report tracks 29 reported Grok 4 benchmark scores across FrontierMath, Berkeley Function Calling Leaderboard, ARC-AGI-1, GPQA Diamond, Humanity's Last Exam, SimpleQA Verified, AIME (OTIS Mock), ARC-AGI-2, and 21 more — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.
29 results
Benchmark results 29
Compare this model| Benchmark | Category | Score | Metric | Status | Date | |
|---|---|---|---|---|---|---|
| FrontierMath | Reasoning | 19.66% | accuracy | Official | Jul 9, 2025 | Details |
| Berkeley Function Calling Leaderboard | Tool use | 62.97% | accuracy | Official | Jul 9, 2025 | Details |
| ARC-AGI-1 | Reasoning | 66.67% | accuracy | Official | Jul 9, 2025 | Details |
| GPQA Diamond | Reasoning | 87.0% | accuracy | Official | Jul 9, 2025 | Details |
| Humanity's Last Exam | Reasoning | 24.52% | accuracy | Official | Jul 9, 2025 | Details |
| SimpleQA Verified | Other | 47.9% | accuracy | Official | Jul 9, 2025 | Details |
| AIME (OTIS Mock) | Reasoning | 84.0% | accuracy | Official | Jul 9, 2025 | Details |
| ARC-AGI-2 | Reasoning | 15.97% | accuracy | Official | Jul 9, 2025 | Details |
| LMArena | Chat preference | 1409 | source-defined rating | Official | Jul 9, 2025 | Details |
| WeirdML | Coding | 45.7% | average accuracy | Official | Jul 9, 2025 | Details |
| Artificial Analysis Intelligence Index | Reasoning | 41.5 | Index | Unverified | Jul 9, 2025 | Details |
| Epoch Capabilities Index | Reasoning | 147.4 | Index | Official | Jul 9, 2025 | Details |
| Aider Polyglot | Coding | 79.6% | % correct | Official | Jul 9, 2025 | Details |
| MMLU-Pro | Reasoning | 86.6% | accuracy | Verified | Jul 9, 2025 | Details |
| GDPval | Agents | 989 | Elo | Official | Jul 9, 2025 | Details |
| LiveCodeBench | Coding | 81.9% | Pass@1 | Unverified | Jul 9, 2025 | Details |
| METR Task-Completion Time Horizons | Agents | 109 min | 50% time horizon | Official | Jul 9, 2025 | Details |
| SciCode | Coding | 45.7% | accuracy | Unverified | Jul 9, 2025 | Details |
| Global-MMLU | Reasoning | 89.5% | accuracy | Unverified | Jul 9, 2025 | Details |
| Search Arena | Chat preference | 1143 | Elo | Verified | Jul 9, 2025 | Details |
| Design Arena | Chat preference | 1070 | Elo | Verified | Jul 9, 2025 | Details |
| MCP-Universe | Tool use | 33.33% | Overall Success Rate | Verified | Jul 9, 2025 | Details |
| OCRBench v2 | Multimodal | 45.0 | accuracy | Verified | Jul 9, 2025 | Details |
| FACTS Grounding | Reasoning | 54.7% | Grounding accuracy | Verified | Jul 9, 2025 | Details |
| ZeroBench | Multimodal | 1.0% (pass@1) | accuracy | Verified | Jul 9, 2025 | Details |
| IMO-Bench | Reasoning | 73.1% | accuracy | Verified | Jul 9, 2025 | Details |
| FrontierMath Tier 4 | Reasoning | 2.1% | accuracy | Official | Jul 9, 2025 | Details |
| Gray Swan Arena (Agent Red-Teaming / Indirect Prompt Injection) | Agents | 2.9% | Attack Success Rate (ASR) | Verified | Jul 9, 2025 | Details |
| Remote Labor Index | Agents | 2.08% | automation rate | Official | Jul 9, 2025 | Details |