---
title: "Claude Opus 5.5 API pricing: ⁨$4⁩ in, ⁨$20⁩ out · DotsAgent"
description: "Claude Opus 5.5 by Anthropic: price per million input, cached and output tokens, context window, agent features and what a real agent task costs on it."
url: https://dotsagent.io/models/claude-opus-5-5
---

[Anthropic](https://dotsagent.io/providers/anthropic)

· Flagship

# Claude Opus 5.5 API pricing and specs

Anthropic's recommended default model, cheaper than Opus 5 was, with adaptive thinking that cannot be switched off and the new computer-use toolset.

Input

$4

Cached input

$0.20

Output

$20

US dollars per million tokens, standard tier, short-prompt rate

Forced tool choice returns an error, so send tool_choice auto with strict tools. Cache writes cost $5 per million for 5 minutes or $8 for an hour. A Fast mode preview costs $8 input and $40 output on the Claude API only.

## Spec sheet

- **API model string**: `claude-opus-5-5`
- **Provider**: [Anthropic](https://dotsagent.io/providers/anthropic)
- **Context**: 1M tokens
- **Max output**: 128K tokens
- **Released**: September 22, 2026
- **Batch API**: 50% off, results within hours
- **Documented agent features**: Tool callingStructured outputComputer useAdjustable reasoning

## What a task costs

Four reference workloads priced at this model's list rates, with caching applied after the first call. Your numbers will differ; the calculator lets you change every assumption.

| Workload | Per task | Per 1,000 tasks |
| --- | --- | --- |
| Support bot 4 calls, short tool lookups | $0.108 | $108 |
| Coding agent 30 calls, files and test output | $3.39 | $3,388 |
| Browser agent 25 calls, screenshots and page text | $2.63 | $2,626 |
| Research agent 15 calls, long search results | $2.55 | $2,548 |

[Change the assumptions](https://dotsagent.io/cost#model=claude-opus-5-5)

## Other Anthropic models

- [Claude Fable 5.1](https://dotsagent.io/models/claude-fable-5-1): $10 / $50

- [Claude Sonnet 5.5](https://dotsagent.io/models/claude-sonnet-5-5): $2 / $10

- [Claude Haiku 4.5](https://dotsagent.io/models/claude-haiku-4-5): $1 / $5

## Similar cost elsewhere

Models from other providers whose cost on the 30-step coding workload is closest to this one.

- [Kimi K3](https://dotsagent.io/models/kimi-k3): $2.97 per coding task

- [Grok 4.7](https://dotsagent.io/models/grok-4-7): $2.68 per coding task

- [Mistral Medium 3.5](https://dotsagent.io/models/mistral-medium-3-5): $5.31 per coding task

- [Gemini 3.1 Pro Preview](https://dotsagent.io/models/gemini-3-1-pro-preview): $2.05 per coding task

## Sources

1. [platform.claude.com](https://platform.claude.com/docs/en/about-claude/models/overview)/docs/en/about-claude/models/overview
2. [platform.claude.com](https://platform.claude.com/docs/en/about-claude/pricing)/docs/en/about-claude/pricing
3. [platform.claude.com](https://platform.claude.com/docs/en/release-notes/overview)/docs/en/release-notes/overview
4. [anthropic.com](https://www.anthropic.com/pricing)/pricing

Independent reference for people who build AI agents. Not affiliated with any vendor named here.

© 2026 DotsAgent · Facts checked October 1, 2026