Summary
Anthropic says Claude Opus 5.5 costs 40% less than Opus 5 on typical workloads at default settings and generates output more than 30% faster. Its per-token prices are also lower, with access controls for biology and cybersecurity work.
Anthropic announced Claude Opus 5.5 on September 22, 2026, calling it the first model in its Claude 5.5 family. The company says the model costs 40% less than Opus 5 on typical workloads at default settings and generates output more than 30% faster. The savings reflect both lower token prices and fewer tokens used per task, according to Anthropic.
Lower prices, with workload savings beyond the token rate
A token is a unit of text that a model processes or generates. Anthropic’s listed prices per million tokens are:
| Token type | Claude Opus 5.5 | Claude Opus 5 |
|---|---|---|
| Input | $4 | $5 |
| Output | $20 | $25 |
| Cache reads | $0.20 | $0.50 |
| Cache writes | $5 | $6.25 |
Input and output prices are 20% lower than Opus 5; cache reads, which Anthropic says account for most costs in agentic and coding work, are 60% lower. The company’s 40% typical-workload estimate is larger than those per-token reductions because Opus 5.5 also uses fewer tokens per task and requires less compute to serve.
A separate fast mode is available in Claude Code and the Claude Platform. Anthropic says it can generate output at up to 2.5 times the speed, at $8 per million input tokens and $40 per million output tokens. The company is also increasing five-hour usage limits for Pro, Max, Team and seat-based Enterprise plans, and says subscribers can save a rate-limit reset for later use.
Benchmark results and safeguards
Anthropic reports improvements across coding, computer-use and knowledge-work evaluations. In its results for Terminal-Bench 4.0, which tests complex multi-step work in a command-line interface, Opus 5.5 scored 66.4%, compared with 52.3% for Opus 5. The Opus 5.5 result used xhigh effort. Anthropic reports a standard error of ±2.6 percentage points for that score and cautions that benchmark margins at this capability level are becoming a less reliable guide to differences in practical work.
The company says Opus 5.5 performed best to date among the models it has tested on its automated behavioral audit, which covers thousands of simulated scenarios. It also says the model is less likely than recent models to take hard-to-reverse actions or exceed its assigned boundaries, and is more resistant to prompt injection than Opus 5. External evaluators, including Frontier Design and METR, tested it before release.
Anthropic says Opus 5.5 has capabilities comparable to Claude Mythos 5.1 in biology and cybersecurity, and is therefore launching with safeguards similar to those used for Claude Fable 5.1. Vetted organisations can apply to use it for biology research through the Life Sciences Verification Program. Anthropic says it plans to expand its Cyber Verification Program in the coming weeks, allowing verified cybersecurity practitioners to use the model for their work.