Claude Opus 5.5 cuts costs, adds banked resets

This is more consequential than a decimal-point upgrade: Anthropic is making frontier-quality agentic work cheaper and easier to sustain. The packaging—lower inference costs, faster output, and more usage headroom—may matter as much as the benchmark gains.

• –API pricing drops to $4 per million input tokens and $20 per million output tokens, while cache reads fall 60% to $0.20 per million tokens.
• –Anthropic claims Opus 5.5 matches Fable 5.1 on most work and leads on reported coding and knowledge-work benchmarks, though independent evaluations remain necessary.
• –Fewer tokens per task plus output generation over 30% faster could compound into significant savings for long-running coding agents.
• –Clearer communication and stronger instruction following should reduce review overhead and developer supervision.
• –Higher limits and banked resets may prove as valuable as the model itself for heavy Claude Code users.