EN

Sonnet 5.5

Claude Sonnet — Anthropic API

$2 input / $10 output per million tokens (Anthropic API)

  • $2 / $10
  • 1M Context
  • Coding
  • Fast

Try Model

Overview

Claude Sonnet 5.5 (API id `claude-sonnet-5-5`) is Anthropic’s default coding and agent model: 1M context, fast latency, and the $2 / $10 price that also applies to Sonnet 5. Use it for daily patches, tool loops, and writing. Escalate to Opus 5.5 or Fable 5.1 only when quality or stakes justify the jump.

Capabilities

  • Daily codingPatches, tests, refactors, and IDE/agent loops where you iterate many times in a session.
  • Cost-aware agentsAt $2 / $10 per million tokens it is half the list price of Opus 5.5 ($4 / $20) and one-fifth of Fable 5.1 ($10 / $50).
  • Prompt cacheCache reads are $0.20 / MTok (10% of input). A stable system prompt plus repo sketch pays for itself after a couple of turns.
  • BatchBatch API is 50% off input and output ($1 / $5) when the job can wait.

Context Window

Sonnet 5.5 is listed at 1M input tokens and up to 128K output. A large window does not mean a cheap window—input is still $2 / MTok, and output is $10 / MTok.

  • Cache the static brief; send only the delta each turn.
  • Trim logs and duplicated files before you burn output tokens on a recap.
  • Confirm the current window on Anthropic’s Sonnet 5.5 model card.

Pricing

Standard API: $2 / MTok in · $10 / MTok out

These are Anthropic list prices for Claude Sonnet 5.5 (same $2 / $10 as Claude Sonnet 5 after the introductory rate was made standard). Hosts (Bedrock, Vertex, OpenRouter, IDEs) may add a markup. MEAGENT Agent Hub uses account credits and is not a Claude invoice.

Input
$2 / million tokens
Output
$10 / million tokens
5-minute cache write
$2.50 / MTok
1-hour cache write
$4 / MTok
Cache read
$0.20 / MTok (0.1× input)
Batch API
$1 / $5 (50% off in/out)
US-only inference
1.1× input and output where offered
  • Worked example: 50K input + 2K output ≈ $0.10 + $0.02 = $0.12 at list (before cache).
  • Cached follow-up on a 50K system prompt: cache read ≈ $0.01 plus new tokens.
  • Sonnet 5.5 is about 30% cheaper to run than Sonnet 5 on typical token mix at the same $2 / $10 sticker, per Anthropic’s product note—verify your own traces.
  • Prices change. Re-check Anthropic’s pricing page before you budget.

Anthropic pricing

How to Use

  1. Pin the SKUUse `claude-sonnet-5-5` (or your host’s equivalent) so you don’t accidentally bill Opus or Fable.
  2. Cache the static prefixPut repo map, style rules, and tool schemas in a cacheable prefix. Reads are $0.20 / MTok.
  3. Watch outputOutput is 5× input. Ask for diffs and short plans, not full-file dumps, unless you need them.
  4. Escalate on purposeIf Sonnet loops, move the recap to Opus 5.5 ($4 / $20) or Fable 5.1 ($10 / $50)—don’t leave the whole day on the expensive SKU.

Alternatives