Anthropic cuts cache reads to $0.25/M tokens — append-only chats save 87% per call
atShruti · x · 2026-09-02
Anthropic dropped cache read pricing to $0.25 per million tokens (a quarter of base). Same call on Claude Fable 5.1 — 200k tokens in, 5k out: $2.25 cold vs $0.30 cached.
- Investor atShruti argues most pre-seed teams optimize model choice, but architecture is the real cost lever
- Practical tip: keep conversations append-only — editing earlier turns invalidates cache and thinking blocks
- Fable 5.1 also ships 3 breaking changes (forced tool use errors, etc.) and beta features like per-message effort and turn-scoped system messages
More from coding & agent
- Developer asks: is LangChain still worth it vs rolling your own agent harness? — curious_vii · 2026-09-03
- Inference Engineering Is Just a Recipe: vLLM/SGLang, Replicas, Cache-Aware Routing — GabGarrett · 2026-09-03
- Developer accidentally built an entire agent factory with Fable 5.1 — 0xkarasy · 2026-09-03
- Microsoft adds Fabric data agents to Foundry agents via Fabric IQ (preview) — adnan_hashmi · 2026-09-03
- Databricks pitches agent-native data infrastructure, Lakebase Postgres at VLDB 2026 — matei_zaharia · 2026-09-03
- doodlestein ships a comprehensive web app review skill after months of debugging — doodlestein · 2026-09-03