Anthropicclosed weightsReasoning
Claude Opus 5
Top of the AA Index at launch; 128K max output.
Jul 24, 2026
$5 / $25
$10
1M
Benchmark scores
| Benchmark | Score | Δ vs prior | Rank (current) | Configuration | Source |
|---|---|---|---|---|---|
Terminal-Bench 2.1 | 84.6% | #4 of 21 | leaderboard | benchlm.ai ↗ | |
SWE-bench Verified | 96% | #1 of 9 | BenchLM ledger; Anthropic did not publish · leaderboard | benchlm.ai ↗ | |
Humanity's Last Exam | 56.6% | #5 of 15 | no tools · lab-reported | vellum.ai ↗ | |
ARC-AGI-2 | 90.4% | #4 of 10 | leaderboard | benchlm.ai ↗ | |
Artificial Analysis Intelligence Index | 60.7 | #1 of 11 | leaderboard | benchlm.ai ↗ | |
LMArena Text | 1505 | #1 of 5 | max · leaderboard | benchlm.ai ↗ |