Anthropic Slashes Claude Sonnet 5.5 Cache-Read Cost by 50%

Anthropic on October 7, 2026, lowered the price of cache reads on Claude Sonnet 5.5 by 50%, from $0.20 to $0.10 per million tokens, effective immediately. The company also introduced a monthly Claude Platform API credit for Max and Team subscribers: $100 per month for Max 5x, $200 per month for Max 20x, and up to $500 per month pooled across users on Team plans.

Both changes were announced alongside the launch of Claude Haiku 5.5, the company’s new small model, in the “Further updates” section of the Claude Haiku 5.5 announcement. Anthropic described the moves as improvements to the value of its models and products.

Sonnet 5.5 Cache-Read Pricing

The reduction applies to cache reads, the tokens billed when a subsequent API request retrieves previously cached content. Cache reads on Sonnet 5.5 now cost $0.10 per million tokens rather than $0.20. Because cache reads make up a large share of models’ token consumption, Anthropic said, the change reduces the cost of running Sonnet 5.5 on most agentic tasks by around 20%. The company illustrated the shift with a Terminal-Bench 4.0 chart plotting Sonnet 5.5’s accuracy against cost per attempt at both the old and new cache-read rates.

The prompt-caching section of Anthropic’s Claude Platform pricing documentation states that a cache hit now costs 5% of the standard input price on Claude Opus 5.5 and Claude Sonnet 5.5: $0.20 per million tokens on Opus 5.5 and $0.10 on Sonnet 5.5. The documentation lists a 2.5% rate on Claude Fable 5.1 and Claude Mythos 5.1 and the standard 10% rate on all other models. Under its multiplier system, five-minute cache writes are billed at 1.25 times the base input price and one-hour writes at twice the base input price. Sonnet 5.5’s remaining rates are unchanged: $2 per million input tokens, $10 per million output tokens, $2.50 per million tokens for five-minute cache writes, and $4 per million tokens for one-hour cache writes.

Anthropic launched Claude Sonnet 5.5 on September 28, 2026, pricing it the same as Sonnet 5 at $2 per million input tokens, $10 per million output tokens, and $0.20 per million tokens for cache reads. The new rate replaces that launch cache-read price while leaving the rest of the model’s pricing intact.

Monthly API Credits by Plan

The second change adds a recurring monthly credit for the Claude Platform, Anthropic’s API environment. Anthropic said the credit is designed to let subscribers experiment with building tools, apps, and agents that call its API, and that it can be used on any of the company’s models. Per the program’s Help Center terms, Max 5x subscribers receive $100 per month and Max 20x subscribers receive $200 per month. On Team plans, credits accrue per seat ($20 per Standard seat and $100 per Premium seat) and pool into a single monthly balance capped at $500. Discounted Team plans, including Nonprofit and Scientists plans, receive the same per-seat amounts. Anthropic’s worked example gives a team with three Standard seats and two Premium seats $260 a month, and states that adding another Premium seat would raise the next cycle’s credits to $360. The pool is calculated from the seats on the plan each time credits are deposited.

Free, Pro, and Enterprise plans are not eligible. A Max or Team subscription must be active and in good standing, and new subscribers can claim once they have been on an eligible plan for seven days. Subscriptions can be purchased on the web, iOS, or Android, but credits are claimed on claude.ai in a web browser, with no payment method required on the Claude Platform. Anthropic’s official developer channel announced the rollout on October 7, 2026, stating that the credits work on any model, including Haiku 5.5, in customers’ own code or third-party harnesses. The Help Center notes the credits are rolling out over a few days and may not appear immediately for every eligible subscriber.

Claiming and Using the Credits

Claiming links a single Claude Console organization to the plan. A Max subscriber, or a Team Primary Owner or Owner, must also hold an Owner, Admin, or Billing role in the Console organization, which can be created during the claim flow. Only one organization can be linked, each organization can receive credits from only one plan, and the linked organization cannot be changed without contacting support.

The credits work with any available Claude model on the Claude Platform: the Messages API and Message Batches API, the Console Playground, Claude Managed Agents, and the Claude Agent SDK. They do not apply to interactive Claude Code sessions in the terminal, IDE, desktop, or web; to extra usage in Claude, Claude Code, or Claude Cowork; or to Claude on Amazon Bedrock, Google Cloud Vertex AI, or Microsoft Foundry.

Credits refresh each billing cycle — monthly even on annual plans — and unused credits expire at the end of each cycle rather than rolling over. The monthly grant is spent before any purchased credits, is shared across everyone holding an API key in the linked organization, and appears in the Claude Console with its amount and expiry date. The credits do not change usage limits in Claude, Claude Code, or Claude Cowork. When the balance runs out, usage draws on purchased credits or auto-reload if the organization has them; otherwise API requests stop until the next monthly grant, and usage is never charged to the subscriber’s Claude plan.

Plan changes carry specific consequences. Canceling, downgrading to an ineligible plan, or refunding stops new credits, while credits already granted stay usable until they expire. Upgrading from Max 5x to Max 20x grants prorated credits immediately, followed by $200 each billing cycle. Moving from Max to Team ends the Max link, and a Team Owner or Primary Owner can claim the team’s credits once the Team plan has been active for seven days. The Help Center states the program is separate from the earlier Agent SDK credit, which is no longer available.