Opus 5.5 cuts cache reads 60%: Addy Osmani breaks down real Claude Code task costs
ClaudeDevs · x · 2026-09-26
claude.dev published a deep dive on what Opus 5.5's price cut means for real Claude Code tasks: 20% cheaper input/output tokens and 60% cheaper cache reads versus Opus 5.
Key points:
- Task cost is driven by turns, cache reads, and output token type — every turn resends the full conversation, so a retry costs more than any token-saving setting.
- Same per-token price can mean very different task costs depending on how many turns/retries a model needs; price tasks, not tokens.
- The post quantifies tradeoffs (lower effort, smaller model, less context) and ships a calculator so you can estimate your own typical task costs from /usage data.
More from coding & agent
- Perplexity CEO shows agent building browser-playable 3D zombie shooter in hours with zero 3D assets — AravSrinivas · 2026-11-11
- The complete harness engineering roadmap: how OpenAI shipped ~1M lines with zero hand-written code — blaizedsouza · 2026-11-11
- Micius-Flash: 27B Multimodal Model for Hardware and Semiconductor Agents — zhiman-ai · 2026-11-11
- Viral 'the harness is the machine' thread touts unverified Claude Fable/Mythos 5.2 9-equation architecture — blaizedsouza · 2026-11-11
- Compaction creates a new failure mode for coding agents: critical bug details can vanish mid-fix — blaizedsouza · 2026-11-11
- What 40+ coding-agent harnesses do between your prompt and the model, dissected — blaizedsouza · 2026-11-11