Grok 4 is a closed-weights model from xAI, released July 9, 2025. It has an ECI score of 146 and ranks 66th of 249 tracked models. At the time of its release, it ranked 3rd of 148 tracked models. At $3/Mtok for input tokens and $15/Mtok for output tokens, it is more expensive per token than similarly-capable models.
| Domain | Benchmark | ||||
|---|---|---|---|---|---|
ECI scoreBest score 166 | 146 | 140 | 147 | 145 | |
| Ranking | #66 | #106 | #57 | #76 | |
| Agent | GDPvalBest score 49.7% | ||||
| Conceptual reasoning | ARC-AGI-2Best score 95.0% | ||||
ARC-AGI-1Best score 98.5% | |||||
| Games | Chess PuzzlesBest score 72.0% | ||||
| Mathematics | OTIS Mock AIME 2024-2025Best score 100.0% | ||||
| Science | GPQA DiamondBest score 95.8% | ||||
| Software engineering | METR Time HorizonsBest score 1044.8 | 110.1 | — | — | 38.7 |
Terminal-Bench 2.0Best score 84.7% | |||||
Aider PolyglotBest score 88.0% |