Latest Models
Scores across benchmarks
Updated Jul 19, 2026, 10:00 UTC
| # | Model | LM Arena ↗ | SWE-bench ↗ | ARC-AGI-2 ↗ |
|---|---|---|---|---|
| 🥇 | Claude Opus 4.6 | 1504 | 75.6% | 69.2%($3.47) |
| 🥈 | GLM 5 | - | 72.8% | 22.8%($0.25) |
| 🥉 | GPT 5.4 | - | - | 83.3%($16.41) |
| 4 | GPT 5.2 | - | 72.8% | 72.9%($38.99) |
| 5 | Gemini 3 Pro | 1486 | 69.6% | 54.0%($30.57) |
| 6 | Claude Opus 4.5 | - | 76.8% | 37.6%($2.40) |



