Anthropic Releases Claude Opus 5.5 With Lower Prices

Anthropic opened the Claude 5.5 family with Opus 5.5, a model the company says matches Claude Fable 5.1 on most work while costing 40 percent less to run than Opus 5. The release follows the company’s public call for pacing frontier development, and Anthropic says Opus 5.5 was tested before launch by external evaluators including Frontier Design and METR. The company frames the model as both a capability step and a cost step, pairing new benchmark scores with a lower per-token price table.

On price, input tokens fall to $4 per million and output tokens to $20 per million, each a 20 percent cut from Opus 5. Cache reads, which Anthropic says make up most agentic and coding costs, drop to $0.20 per million tokens, a 60 percent cut. Cache writes fall to $5 per million from $6.25. Alongside the price changes, Anthropic is raising five-hour usage limits on Pro, Max, Team and seat-based Enterprise subscription plans, and adding a rate limit reset that subscribers can save and apply when they choose.

On capability, Anthropic reports gains on agentic coding and knowledge work benchmarks, including Terminal-Bench 4.0 and GDPval-AA v2.1, and says Opus 5.5 scored higher on its internal alignment suite than any model it has tested to date. The company is deploying Opus 5.5 with cybersecurity and biology safeguards similar to those on Fable 5.1, and it is expanding its Cyber Verification Program and Life Sciences Verification Program to cover the new model. Preserved thinking, the anti-distillation measure introduced with Fable 5.1, also applies to Opus 5.5 for eligible API accounts.

The numbers

Anthropic’s own pricing table sets Opus 5.5 at $4 input and $20 output per million tokens, against $5 and $25 for Opus 5, with cache reads cut to $0.20 from $0.50. The company’s headline figure is a 40 percent drop in typical workload cost, a claim it attributes to lower per-token pricing combined with fewer tokens used per task rather than price alone.

Fast mode for Opus 5.5, available in Claude Code and the Claude Platform, costs $8 per million input tokens and $40 per million output tokens, roughly double the standard rate for faster responses. Anthropic did not publish separate pricing for the Sonnet 5.5 and Haiku 5.5 models it says will follow in coming weeks, so the near-term comparison for buyers is between Opus 5.5 and the existing Sonnet 5 and Haiku 4.5 lines already listed on the company’s pricing page. Anyone budgeting for a coding agent that leans on cached context will feel the $0.20 cache read rate more than the headline output price, since cache reads carry most of the token volume in long sessions.