Performance and Efficiency
Anthropic reports that Opus 5.5 is a "major step up from Opus 5" and serves as the company's new leading model. In internal testing, the company stated that one tester completed a 680,000-line code migration in less than a day, a task they claim would have taken an engineering team weeks.
On specific benchmarks, Anthropic reports the following for Opus 5.5:
- Agentic coding (Terminal-Bench 4.0): 66.4%
- Knowledge work (GDPval-AA v2.1): 1846 Elo
- Computer use (OSWorld 2.1): 81.8% (partial)
Anthropic notes that while Opus 5.5 leads in several categories, including agentic coding and knowledge work, "benchmark margins have become a less reliable guide to real-world differences."