| Claude Opus 4.7 | Claude Opus 4.6 | |
|---|---|---|
| arenaElo | 1420 | 1402 |
| humanEval | 93.1 | 91.4 |
| sweBench | 64.3 | 53.4 |
| mmlu | 92.3 | 90.8 |
| gpqa | 65.2 | 62.1 |
| aime | 88 | 84 |
| Price in/out ($/M) | $15 / $75 | $15 / $75 |
Opus 4.7 doubles the context window (1M vs 500K), improves GPQA by ~3 points and pushes HumanEval past 93%. Pricing is identical. If you already run on 4.6 and aren't hitting context limits, 4.7 is a drop-in upgrade worth testing; otherwise migrate for the long-context wins.
Claude Opus 4.7 · Claude Opus 4.6 · All AI model comparisons