Grok 4 Fast (Reasoning): benchmark results
Grok 4 Fast evaluated with reasoning enabled. Provider: xAI. Released 2025-09-19. Access: API.
Unified ELO 1599 ± 1, rank #440 of 1761 rated models, from 154 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| NYT Connections Older Models | 88 | Score (%) | 96.8 |
| LiveOIBench | 69.8 | Avg Human Percentile | 96.5 |
| AA LiveCodeBench | 83.17 | Pass@1 (%) | 94.4 |
| LLM2014 Logic 2025-09 | 67.59 | Median Score | 93.3 |
| UGI Leaderboard | 51.29 | UGI Score | 92.1 |
| LLM2014 Logic 2025-10 | 67.76 | Median Score | 91.8 |
| Vals AI TaxEval v2 | 75.7 | Accuracy (%) | 91.7 |
| Diplomacy: Overall Performance | 57.75 | Score | 90.9 |
| AA AIME 2025 | 89.67 | Accuracy (%) | 90.7 |
| AA MMLU-Pro | 85.04 | Accuracy (%) | 89.8 |
| UGI - Writing | 53.37 | Writing Score | 89.8 |
| UGI - Natural Intelligence | 46.53 | NatInt Score | 88.8 |
Interactive version: theaggregate.ai/model?slug=grok-4-fast-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-05.