Anthropic Debuts Claude Opus 5.5, Matching Fable 5.1 Capabilities at 40% Lower Run Costs

Anthropic has officially launched Claude Opus 5.5, the first release in its new Claude 5.5 model family. Positioned directly between flagship frontier reasoning and cost-effective production deployment, Opus 5.5 performs at the level of Claude Fable 5.1 across the majority of enterprise workloads while reducing runtime operating costs by 40% compared to Opus 5 at default settings.
The launch establishes significant per-token and infrastructure-level cost reductions across Anthropic's developer API:
- Token Pricing: Listed at $4.00 per million input tokens and $20.00 per million output tokens—representing a direct 20% cut compared to Opus 5 ($5/$25) and a 60% discount against Fable 5.1 ($10/$50).
- Prompt Caching Slashed: Cache read rates have been cut 60% to $0.20 per million tokens (down from $0.50), significantly reducing inference spend on repetitive, multi-turn agentic coding and document-processing loops.
- Throughput and Defaults: Token generation speeds have increased by more than 30% over Opus 5, with the API defaulting to medium effort rather than high.
On benchmark evaluations, Opus 5.5 posts substantial gains in software engineering and professional knowledge work—scoring 66.4% on Terminal-Bench 4.0 and 54.4% on FrontierCode v1.1—while narrowing the gap with external frontier models like OpenAI's GPT-6 Astra. Anthropic is deploying the model with enhanced safety guardrails matching Fable 5.1, requiring vetted clearance via its Life Sciences Verification Program for biology research alongside strict cybersecurity protections.
The release intensifies margin and unit-economic pressure across frontier AI providers, accelerating enterprise adoption by sharply cutting the cost curve for complex agentic workflows and multi-step codebase refactoring.