Claude Code Cost Figures Now Reflect Cheaper Sonnet 5.5 Cache Reads
Claude CodeClaude Code 2.1.296 updates /cost, the status line, --max-budget-usd and the SDK's cost figures to price Claude Sonnet 5.5 cache reads at $0.10 per million tokens, down from $0.20. Because Claude Code leans heavily on prompt caching in long sessions, the lower rate shows up directly in reported spend and in how far a budget cap stretches.
Key Takeaways
- Sonnet 5.5 cache reads are now priced at $0.10 per million tokens in Claude Code, down from $0.20.
- The new rate flows into
/cost, the status line,--max-budget-usdand the SDK's cost figures. - Cache reads are usually the largest cost in long Claude Code sessions, so halving the rate has an outsized effect on total spend.
--max-budget-usdcaps stretch further because spend accrues more slowly under the lower rate.- Third-party trackers tie the change to an Anthropic price cut announced on October 7, 2026, with cache writes and other prices unchanged.
- Anthropic's pricing page was reported to show mixed figures, so the current page should be checked before relying on the number.
Sources & Mentions
2 external resources covering this update
A lower cache-read rate for Sonnet 5.5
Claude Code 2.1.296 changes how it prices Claude Sonnet 5.5 cache reads. The figure is now $0.10 per million tokens, where it was $0.20 before. The update applies everywhere Claude Code reports or enforces cost: the /cost command, the status line, the --max-budget-usd limit and the cost figures exposed through the Agent SDK.
Third-party trackers report that Anthropic announced the matching price cut for Sonnet 5.5 prompt-cache reads on October 7, 2026, halving the rate while leaving cache writes and other prices unchanged. One of those trackers also notes that the summary table on Anthropic's pricing page had not been updated to the new figure at the time of writing, while the prompt-caching section had, so readers should check the current pricing page before budgeting.
Why cache reads dominate in Claude Code
A long Claude Code session resends a large, mostly unchanged context on every turn: the system prompt, tool definitions, project instructions and conversation history. Most of that input is served from the prompt cache, so cache reads are often the largest line in the token bill for agentic work. Halving that rate changes the cost of a long session more than a change to the headline input price would.
Effects on budgets
Since --max-budget-usd compares spend against a cap, a session running on Sonnet 5.5 now accrues cost more slowly in Claude Code's accounting, and the same cap allows more work before it triggers. Anyone who recorded per-session costs before this release should expect Sonnet 5.5 sessions to show lower figures for the same activity.