| Claude Opus 4.8 | Claude Opus 4.7 | |
|---|---|---|
| arenaElo | 1435 | 1420 |
| humanEval | 94.5 | 93.1 |
| sweBench | 67 | 64.3 |
| mmlu | 93 | 92.3 |
| gpqa | 67.5 | 65.2 |
| aime | 90 | 88 |
| Price in/out ($/M) | $5 / $25 | $5 / $25 |
Opus 4.8 is a clean upgrade at the same list price, agentic coding rises from 64.3% to 69.2% on SWE-Bench Pro, it leads computer use at 84% on Online-Mind2Web, runs autonomously for longer and reports progress more honestly. Fast mode is now 3× cheaper and the minimum cacheable prompt drops to 1,024 tokens. There is no reason to start a new project on 4.7, go straight to 4.8.
Claude Opus 4.8 · Claude Opus 4.7 · All AI model comparisons