Sonnet 5.5 cache reads drop to $0.10 per million

Cache reads on Claude Sonnet 5.5 are cheaper on the Claude Platform API. ClaudeDevs said on October 8, 2026 that the price has been cut in half, to $0.10 per million tokens. Input is still $2 per million, and output is still $10 per million. The post said the change is API only, with no change to Claude Code usage limits.

ClaudeDevs wrote that this "means Sonnet 5.5 now runs ~20% cheaper on most agentic work." That is the account's statement. The post does not include a workload sample or an independent cost measurement.

What the model page lists

The Claude Platform Sonnet 5.5 model page lists the same cache-read price. Current rates on that page:

Input: $2 per million tokens
Output: $10 per million tokens
Five-minute cache write: $2.50 per million tokens
One-hour cache write: $4 per million tokens
Cache read: $0.10 per million tokens

The what's-new page says Sonnet 5.5 has the same prices as Claude Sonnet 5 except for prompt cache reads, which cost $0.10 USD per million tokens, half the Claude Sonnet 5 rate. Set against the listed $2 input price, a cache read is 5 percent of base input. The page does not describe a cut to cache writes.

The model page still shows the headline comparison rate as $2 / $10 per million tokens. That pair is input and output, not the cache-read line. Sonnet 5.5, API ID claude-sonnet-5-5, was released September 28, 2026, with a 1 million token context window and a 128K max output. The October 8 post does not give a separate effective date for the price change, and the docs do not date it on their own. The minimum cacheable prompt length listed for this model is 512 tokens.

If you bill agent work on the Claude Platform API, price Sonnet 5.5 cache-read tokens at $0.10 per million and leave Claude Code usage limits out of the comparison. Read the pricing block on the Sonnet 5.5 model page and the pricing note on the what's-new page before treating write rates as part of the cut.

What the model page lists

Sources