Anthropic has introduced Claude Opus 5.5, the first model in its new Claude 5.5 family. According to the company, Opus 5.5 performs at the level of Claude Fable 5.1 on most work while costing 40% less to run than Opus 5.
The model was tested by external evaluators, including Frontier Design and METR, before release. On Anthropic’s automated behavioral audit—described as their most comprehensive alignment test—Opus 5.5 achieved the strongest performance of any model tested to date.
Performance Improvements
Opus 5.5 represents a significant step up from Opus 5. Early testers reported large performance jumps on complex work. One tester completed a 680,000-line code migration in less than a day, work that would have taken an engineering team weeks. In web performance optimization, Opus 5.5 succeeded 39 of 40 times when tasked with cutting load times across every page of an app, while Opus 5 made smaller improvements but also altered the app’s behavior.
The model is particularly strong at agentic coding tasks. On benchmarks, it leads in agentic coding, computer use, and knowledge work. In an internal test, Opus 5.5 translated HAProxy—widely used software that balances web traffic—from C into Rust in 9.5 hours, compared to 12 hours for Fable 5.1, while costing 51% less.
Cost and Speed
Opus 5.5 requires less compute to serve than Opus 5. Input and output tokens cost $4 and $20 per million respectively—20% less than Opus 5. Cache reads cost $0.20 per million tokens, 60% less than Opus 5. The model also generates output more than 30% faster than its predecessor.
Anthropric is increasing five-hour usage limits on Pro, Max, Team, and seat-based Enterprise plans, and providing subscription users a rate limit reset they can save and use at their discretion.
Safety and Communication
Opus 5.5 is less likely than recent models to take hard-to-reverse actions or act outside its boundaries, and shows greater resistance to prompt injection than Opus 5. The model communicates more naturally than prior versions, with early testers noting clearer, easier-to-follow writing that prioritizes important information first.
Because Opus 5.5 is comparable to Claude Mythos 5.1 in biology and cybersecurity, it is being deployed with similar safeguards. Vetted organizations can apply to Anthropic’s Life Sciences Verification Program to use Opus 5.5 for biology research, and cybersecurity verification programs are being expanded in coming weeks.
Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks with similar improvements to performance, efficiency, and safety.
Key facts
- Opus 5.5 costs 40% less than Opus 5 while matching performance of Claude Fable 5.1
- Input and output tokens are 20% cheaper; cache reads are 60% cheaper than Opus 5
- Output generation is 30% faster than Opus 5
- Early tester completed 680,000-line code migration in less than a day
- External evaluators including Frontier Design and METR tested the model before release
- Model achieved strongest performance on Anthropic’s automated behavioral audit