Claude Opus 5.5

anthropicanthropic/claude-opus-5-5

Anthropic's Opus-tier model released 2026-09-22 — successor to Claude Opus 5 for long-running agentic coding and knowledge work; adaptive thinking (always on) with effort control; 1M context, 128K output.

Claude Opus 5.5 model page ↗What's new in Claude Opus 5.5 ↗Claude pricing documentation ↗

Claude Opus 5.5 is Anthropic's Opus-tier model and the successor to Claude Opus 5. Anthropic describes it as built "for long-running agentic coding and knowledge work" and recommends it as the starting point for most workloads. In TheRouter it is cataloged as anthropic/claude-opus-5-5, with text and image input, text output, a 1M-token context window and up to 128K output tokens.

The list price is $4 / $20 per million input / output tokens, below Claude Opus 5's $5 / $25. The API contract changes: Anthropic lists four breaking changes for code already running on Opus 5. The two that most often break existing clients are that thinking can no longer be disabled or given a manual token budget (both return a 400), and that forced tool use (tool_choice any / tool) returns a 400.

Best for

• Long-running agentic coding and knowledge work, which is the workload Anthropic positions this model for
• Long-context work up to 1M tokens, billed at the standard rate across the whole window

Reach for something else if

• Clients that force tool calls with tool_choice: any / tool (or OpenAI required / a named function) — these return a 400; keep auto and use strict tool use, or stay on Claude Opus 5 until migrated
• Workloads that need thinking switched off or a fixed thinking budget — Anthropic rejects both with a 400 on this model; lower the effort level instead
• Latency- and cost-sensitive simple chat — route to Claude Sonnet 5.5 or Claude Haiku 4.5

Context Length

Max Output

128K

Input Priceper 1M tokens

$4.32/ 1M tokens

Output Priceper 1M tokens

$21.60/ 1M tokens

Modalities

textimage→text

Capabilities

Vision

Pricing Breakdown

Type	Rate
Input	$4.32 / 1M tokens
Output	$21.60 / 1M tokens

Supported Parameters

temperaturemax_tokenstop_ptoolstool_choiceresponse_formatreasoningstop

Specifications

Anthropic model id	claude-opus-5-5platform.claude.com ↗	verified
Context window	1,000,000 tokensplatform.claude.com ↗	verified
Maximum output	128,000 tokensplatform.claude.com ↗	verified
Thinking	Adaptive, always on; default effort `medium`platform.claude.com ↗	verified
Reliable knowledge cutoff	June 2026platform.claude.com ↗	verified
License	Anthropic Usage Policy (proprietary, API-only)www.anthropic.com ↗	verified
Pricing (input / output)	$4 / $20 per MTok list price; cache read $0.20, 5-minute cache write $5, 1-hour cache write $8platform.claude.com ↗	verified

Benchmarks

Benchmark	Distribution	Score	Source
Official benchmark table The cited model documentation publishes no benchmark table at curation time.	—	Not publicly disclosed	—

API Usage Examples

Use the global api.therouter.ai endpoint shown below for new integrations; the legacy China accelerated endpoint is retired.

cURL

curl https://api.therouter.ai/v1/chat/completions   -H "Content-Type: application/json"   -H "Authorization: Bearer $THE_ROUTER_API_KEY"   -d '{
    "model": "anthropic/claude-opus-5-5",
    "messages": [
      {"role": "user", "content": "Summarize the key points from this input."}
    ]
  }'

API guide

Chat completion

Use the OpenAI-compatible chat completions endpoint. Leave tool_choice on auto.

cURL

curl https://api.therouter.ai/v1/chat/completions \
  -H "Authorization: Bearer $THEROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"anthropic/claude-opus-5-5","messages":[{"role":"user","content":"Hello"}]}'

More from anthropic

anthropic/claude-opus-5Predecessor model anthropic/claude-sonnet-5-5Faster, cheaper Claude route from the same generation anthropic/claude-fable-5-1Higher-capability Claude route for demanding reasoning

Similar models

Cross-provider sibling models

openai/gpt-5.6-sol

Comparable general-purpose reasoning route

Frequently asked

Can I disable thinking or set budget_tokens on Claude Opus 5.5?

No. Anthropic returns a 400 for thinking.type disabled and for a manual budget_tokens on this model. Thinking is always on; the effort level is the control for depth, latency and cost.

re-authored by TheRouter

Does Claude Opus 5.5 support tool_choice: any or tool_choice: tool?

No. Anthropic rejects both with a 400, and TheRouter rejects them before any upstream call. Use tool_choice: auto with strict tool use, or structured outputs.

re-authored by TheRouter

Fact ledger — every claim on this page traces here

source	URL	retrieved
Anthropic model id	platform.claude.com ↗	2026-10-10	verified
Context window	platform.claude.com ↗	2026-10-10	verified
Maximum output	platform.claude.com ↗	2026-10-10	verified
Thinking	platform.claude.com ↗	2026-10-10	verified
Reliable knowledge cutoff	platform.claude.com ↗	2026-10-10	verified
License	www.anthropic.com ↗	2026-10-10	verified
Pricing (input / output)	platform.claude.com ↗	2026-10-10	verified
Official benchmark table	platform.claude.com ↗	2026-10-10	unknown