Claude Opus 5.5 launches with lower API prices

Quick answer

Anthropic released Claude Opus 5.5 on September 22, 2026. Standard API input and output tokens cost 20% less than Opus 5, while Anthropic estimates a 40% saving on typical workloads after accounting for efficiency. Those are different claims, and neither promises a cheaper app subscription. Claude also gets higher five-hour limits on Pro, Max, and Team plans. This is a source-based explainer, not a hands-on comparison.

Download Chat AI Opens the official App Store or Google Play for your device.

Where can you use Claude Opus 5.5?

Anthropic announces availability across its platforms and AWS, Google Cloud, and Microsoft Azure. The Claude API identifier is claude-opus-5-5. In Claude's chat interface, open the model menu beside the send button and check More models if needed; Enterprise administrators can restrict model access.

Sources: Anthropic, Anthropic, Anthropic Help Center

API prices: 20% lower token rates, not a guaranteed 40% discount

Anthropic's standard US-dollar rates per million tokens are listed below. Cache reads reuse previously processed input; creating the cache has a separate charge.

Uncached input: $4, down from $5 for Opus 5.
Output: $20, down from $25.
Cache reads: $0.20, down from $0.50.
Cache writes: $5 for five-minute caching or $8 for one-hour caching.

Sources: Anthropic, Anthropic

Higher session limits and a reset you choose when to use

The launch raises five-hour usage limits on Pro, Max, and Team, without specifying a universal message allowance. For an available reset, inspect Settings > Usage on Claude's website or desktop app before redeeming it.

Check which allowance the offer restores: a five-hour session limit or weekly usage. The normal weekly reset schedule stays unchanged.
Check the displayed expiry. Redeeming a reset cannot be undone, and it does not refund extra usage already billed.
The reset button is not currently offered in Claude Mobile or the Claude Code terminal or IDE. A reset redeemed on web or desktop applies to the shared account limits.

Sources: Anthropic, Anthropic Help Center

Thinking stays on, and some requests switch models

Opus 5.5 uses always-on adaptive thinking, with medium as its default effort in the API documentation. Higher effort can consume more tokens and reach usage limits sooner. Changing the model name in an existing integration is not a complete migration: Anthropic documents breaking changes, including errors for forced tool use.

Sources: Anthropic, Anthropic Help Center, Anthropic Help Center

Should you change your everyday AI workflow?

Anthropic reports improvements in writing, coding, and efficiency. We have not independently tested those gains. Before switching, use a non-sensitive task whose answer you can check: for example, ask both your current model and Opus 5.5 to extract decisions, owners, and unresolved questions from the same short document.

Sources: Anthropic

Opus 5.5 availability in Chat AI is not verified

Chat AI's published catalog does not list Claude Opus 5.5 as an exact entry. Do not download Chat AI expecting access to this release. If you are choosing a multi-model app, use the comparison below to check model versions, document tools, and limits against your own needs.

Sources: Chat AI

Frequently asked questions

What readers usually ask

Does the price reduction make a Claude subscription 40% cheaper?

No. The 40% figure is Anthropic's estimated API workload saving, not an announced subscription discount. Standard input and output token prices fall by 20%.

Can I turn thinking off for Opus 5.5?

No. Adaptive thinking is always on, including through the API. Use effort settings to adjust how much thinking it applies instead.

Why does my answer say Opus 4.8 or Opus 5 instead?

Claude's safeguards can rerun certain flagged requests on another model when automatic switching is enabled. Read the response label and model picker before assuming Opus 5.5 produced the answer.

Evidence