| Claude Opus 5 | Claude Fable 5 | |
|---|---|---|
| arenaElo | 1492 | |
| humanEval | 98 | |
| sweBench | 97 | 77 |
| mmlu | 95.1 | |
| gpqa | 92 | |
| aime | 94 | |
| Price in/out ($/M) | $5 / $25 | $10 / $50 |
Opus 5 takes the overall lead: 43.3% on Frontier-Bench v0.1 vs 33.7% for Fable 5, a GDPval-AA v2 Elo of 1861 vs 1747, 70.6% vs 66.1% on OSWorld 2.0 and an Artificial Analysis Intelligence Index of 61 at max effort, the highest in the world. It does all of this at $5 / $25 per M tokens, half of Fable 5's $10 / $50. Fable 5 keeps a narrow edge on DeepSWE v1.1 (69.7% vs 68.8%), but for coding, agentic knowledge work and computer use, Opus 5 is the new default. Benchmark both on your real workload before committing.