Claude Sonnet 5.5 vs Claude Sonnet 5

Specs side by side

Price per token

Prices exclude taxes and are charged from your prepaid USD balance. Both models work through the same key, in the OpenAI-compatible /v1/chat/completions endpoint. Failed generations cost nothing.

For new projects pick Claude Sonnet 5.5: it has the same context window and output limit as Sonnet 5, and on Jantaris it costs less. Keep Sonnet 5 where your code relies on disabled thinking or forced tool choice - on Sonnet 5.5 both end in a 400 error and need a small rewrite.

Key differences

• Price. Claude Sonnet 5.5 costs $1.89 / $9.45 per 1M tokens, Claude Sonnet 5 costs $3.24 / $16.20 per 1M tokens. Input and output are both cheaper on the newer version.
• Thinking. On Sonnet 5 you can send thinking: {"type": "disabled"}. On Sonnet 5.5 that is a 400 error - lower effort instead, or send thinking: {"type": "between_tools"} through /v1/messages (at high effort or below).
• Tools. Sonnet 5.5 rejects forced tool choice (tool_choice of type any or a named tool). auto with an instruction in the prompt works.
• Effort. Both default to high, but Anthropic recalibrated the levels in 5.5. The same effort can produce a different amount of reasoning.
• Shared. 1,000,000-token window, up to 128K output tokens, both endpoints: /v1/chat/completions and /v1/messages, and non-default temperature, top_p and top_k are rejected.

When to pick Claude Sonnet 5.5

• New code and new agents written for tool_choice: auto from the start.
• The default model in Claude Code when you want a lower per-token rate.
• Long documents and repositories that fit in the 1M-token window.

When to stay on Claude Sonnet 5

• Production code with thinking disabled that you don't want to touch right now.
• Pipelines that force a specific tool, for example to extract data in a fixed format.
• effort settings tuned for Sonnet 5, until you have re-tested them on 5.5.

Cost in practice

For the same token count Sonnet 5.5 comes out cheaper, since both rates are lower. Thinking is billed as output, so the real cost depends mostly on effort: start at low for chat and simple tasks, medium for agents and code. Current rates for every Claude model are on the pricing page.

FAQ

Is Sonnet 5.5 better than Sonnet 5?

It is the successor in the same line, with the same window and output limit. I don't have my own head-to-head tests yet, so check a few of your own requests on both before switching production.

How do I move code from Sonnet 5 to 5.5?

Remove thinking: {"type": "disabled"} and forced tool_choice, and use a lower effort and tool_choice: auto instead. The rest of the request stays the same, only the model id changes.

Can I use both models in Claude Code?

Yes. Set one as the default and switch with /model during a session. Setup is in the Claude Code guide.