CoreCodeBlog

Models

Claude Sonnet 5.5: pricing, specs and how to use it in Claude Code

Published · Last checked

Claude Sonnet 5.5 was released on September 28, 2026, and Anthropic calls it the best combination of speed and intelligence. It costs half as much as Opus 5.5, runs faster, and is the usual model for everyday coding in Claude Code. This post collects its official specs and prices, how it differs from Sonnet 5 and Opus 5.5, and how to use it in Claude Code. Specs and prices come from Anthropic's Sonnet 5.5 model page and pricing page, checked on November 2, 2026.

Key points
  • The model ID is claude-sonnet-5-5, with a 1M-token context window, up to 128K output tokens, and a June 2026 knowledge cutoff.
  • It costs $2 per million input tokens and $10 per million output tokens, the same as Sonnet 5. Cache reads cost 0.05 times the input price, $0.10, half of Sonnet 5's.
  • Adaptive thinking is on by default and between_tools turns off up-front thinking; the default effort is high on the API and medium in Claude Code.
  • Moving from Sonnet 5 has five breaking changes, and text between tool calls now comes back in thinking blocks.
  • In Claude Code, select it with /model claude-sonnet-5-5; its context is 1M natively, so sonnet[1m] is not needed.

Official specs

ItemClaude Sonnet 5.5
Model IDclaude-sonnet-5-5 (anthropic.claude-sonnet-5-5 on Bedrock)
ReleasedSeptember 28, 2026; retires no sooner than September 28, 2027
Context window1M tokens
Max output128K tokens (300K on the Batches API with a beta header)
ThinkingAdaptive, on by default
Default efforthigh on the API, medium in Claude Code
Comparative latencyFast
Input → outputText and images → text
Reliable knowledge cutoffJune 2026

Pricing

ItemPer million tokens
Input$2
5-minute cache write$2.50
1-hour cache write$4
Cache read$0.10 (input × 0.05)
Output$10
Batch API50% off input and output: $1 / $5

Sonnet 5.5 costs half of Opus 5.5 and a fifth of Fable 5.1 per token. Like Opus 5.5 it reads the cache at 0.05 times the input price, so a long Claude Code session, where most tokens come from the cache, stays cheap. How one request is billed is worked through in Claude API pricing.

Compared with Sonnet 5 and Opus 5.5

Sonnet 5.5Sonnet 5Opus 5.5
Input / output$2 / $10$2 / $10$4 / $20
Cache read$0.10$0.20$0.20
Comparative latencyFast—Moderate
Can thinking be turned offUp-front thinking, yes—No

Code that calls the API directly and moves from Sonnet 5 has five breaking changes (What's new in Claude Sonnet 5.5):

One more change alters the response without failing any request: text between tool calls comes back in thinking blocks. An application that shows that text as progress has to set display, or turn off up-front thinking with between_tools. In Claude Code, Claude Code handles all of this.

When to use Sonnet 5.5

Start everyday coding on Sonnet 5.5: features, bug fixes, tests and single-module refactors. It is half the price and faster. Move to Opus 5.5 for changes across many files or design problems that need sustained reasoning, and to Fable 5.1 for the hardest, longest tasks. opusplan lets Opus plan and Sonnet execute. More advice is in Choosing a Claude model.

Using it in Claude Code

On the Anthropic API, Claude Code's sonnet alias resolves to Sonnet 5.5. Aliases move with new releases, so to pin the version, or when using a relay, name the model in full:

/model claude-sonnet-5-5

To start every session on Sonnet, set it in ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://corecode.live",
    "ANTHROPIC_AUTH_TOKEN": "sk-your-key"
  },
  "model": "claude-sonnet-5-5"
}

Sonnet 5.5 has a native 1M context window, so sonnet[1m] adds nothing. Claude Code runs it at medium effort; use /effort high for harder tasks. Pick the model when the task starts: switching mid-session rewrites the cache at the write price on the next turn.

Safety classifiers and automatic fallback

Sonnet 5.5 runs with safety classifiers. Claude Code re-runs a cybersecurity-flagged request on Sonnet 5; a biology-flagged request has no fallback model and ends with a refusal. CoreCode also serves Sonnet 5, so a cybersecurity fallback runs as it does on the Anthropic API, and the re-run request is billed at Sonnet 5's price.

Using it through CoreCode

Both of CoreCode's Claude groups serve claude-sonnet-5-5, answered by official Anthropic Claude accounts, with each session kept on one account so the cache keeps hitting. The groups are compared in Groups, prices are in the price table, and setup is in Getting started.

FAQ

What does Sonnet 5.5 improve over Sonnet 5?

Same list price ($2 / $10), with cache reads down from $0.20 to $0.10, which makes long Claude Code sessions cheaper. In exchange there are five breaking API changes to check in code that calls the API directly.

Sonnet 5.5 or Opus 5.5?

Sonnet 5.5 for everyday coding, at half the price and faster; Opus 5.5 for complex reasoning and large changes. opusplan splits the work between them.

Can I turn off Sonnet 5.5's thinking?

You can turn off up-front thinking with thinking: {type: "between_tools"} at high effort or below. That is an API setting; in Claude Code, adjust effort with /effort.

Do I still need sonnet[1m] in Claude Code?

No. Sonnet 5.5 has a 1M context window natively, so sonnet[1m] adds nothing.

Sources