Anthropic Launches Claude Opus 5.5 With 40% Lower Costs
The new Opus model targets Fable 5.1-level performance while cutting typical workload costs and improving speed.
Anthropic has launched Claude Opus 5.5, the first model in its new Claude 5.5 family. The company says Claude Opus 5.5 reaches the level of Claude Fable 5.1 for most tasks while costing 40% less to run than Opus 5 on typical workloads.
The release arrives about two months after Opus 5. Anthropic is positioning the new model around long-running coding, agentic work, and professional tasks, with lower token prices and faster output.
Claude Opus 5.5 Cuts Costs While Raising Performance
Anthropic says Opus 5.5 uses less compute to serve than Opus 5. Input tokens now cost $4 per million and output tokens cost $20 per million, down from $5 and $25. Cache reads fall to $0.20 per million tokens, a 60% reduction from Opus 5.
Those token-price changes are only part of the savings. Anthropic says its testing shows typical workloads cost about 40% less overall, while the model generates output more than 30% faster than Opus 5.
For readers tracking Anthropic's model pricing, the existing Claude Pricing 2026 guide provides broader context on Claude's plans and API rates. Saganote also covered Claude Sonnet 5's agentic tool use and pricing, while Claude Sonnet 5 vs Opus 4.8 provides earlier context on the two main Claude tiers.
Long Coding Tasks Are a Major Focus
Anthropic is also emphasizing work that requires an agent to operate across a large codebase for an extended period. In one early test, Opus 5.5 audited and fixed a 200,000-line codebase in under three hours. Anthropic says Opus 5 took more than 20 hours and used 2.5 times as many tokens for the same task.
In another internal test, Opus 5.5 translated HAProxy from C to Rust in 9.5 hours, compared with 12 hours for Fable 5.1, while costing 51% less. Anthropic also says Opus 5.5 beat GPT-6 Astra at roughly 20% of the cost per task on FrontierCode and matched Astra at about 40% of the cost on Terminal-Bench 4.0.
That focus builds on Anthropic's earlier Opus releases. Saganote's coverage of Claude Opus 5 tracks the previous generation's coding and cost positioning, while Claude Opus 5 vs Fable 5 compares the previous flagship models across benchmarks and pricing.
The broader Claude lineup has also expanded in recent months, including Claude Fable 5.1 and Mythos 5.1. Saganote has also tracked changes around Claude Fable 5 access and the broader ChatGPT vs Claude pricing picture.
Safety Testing Expands With Opus 5.5
Anthropic says Opus 5.5 was evaluated by external groups including METR and Frontier Design before release. On Anthropic's most comprehensive alignment test, the company says the model achieved its strongest score to date.
Anthropic also reports that Opus 5.5 was about 85% less likely than Opus 5 or Mythos 5.1 to attempt to bypass containment boundaries in a dedicated internal evaluation. Cybersecurity and biology requests that trigger its safeguards can be routed to other Claude models under Anthropic's deployment controls.
The model is available now through Claude and the Claude API, as well as major cloud platforms including Amazon Web Services, Google Cloud, and Microsoft Azure. Anthropic says Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks with many of the same performance, efficiency, and safety improvements.
Anthropic is also building Claude beyond general-purpose model releases. Recent Saganote coverage includes Claude Science, while the company's model competition can be seen in comparisons such as Grok 4.5 vs Claude Fable 5 vs GPT-5.5.