Claude Opus 5.5 API

Price check: official, OpenRouter and Wokey

USD per million tokens. Official and OpenRouter rates were checked by hand on 2026-09-29, excluding tax; the Wokey rate is read from the live price list.

Checked 2026-09-29 · Official pricing page · OpenRouter endpoint list

Use it in Claude Code

This model supports /v1/messages on Wokey. Set these variables, then start claude.

export ANTHROPIC_BASE_URL="https://api.wokey.ai"
export ANTHROPIC_AUTH_TOKEN="$WOKEY_API_KEY"
export ANTHROPIC_MODEL="claude-opus-5-5"

Claude Opus 5.5 at a glance

Release date

2026-09-22, the first Claude 5.5 model; Sonnet 5.5 followed on 09-28

Positioning

Built for long-running agentic coding and knowledge work. Anthropic says it performs at the level of Fable 5.1 on most work, costs 40% less to run than Opus 5 and outputs more than 30% faster

Context and output

1M context with no beta header; 128K max output, or 300K on the Batch API with the output-300k-2026-03-24 beta header

Input and cutoff

Text and image input; knowledge cutoff June 2026

Thinking and effort

Adaptive thinking is always on and cannot be disabled; five effort levels (low / medium / high / xhigh / max), with medium as the API default (Opus 5 defaults to high)

Official price

$4 input / $20 output per million tokens (Opus 5: $5 / $25); cache read $0.20, 5% of input (Opus 5: $0.50); 5-minute cache write $5, 1-hour $8; Batch 50% off

Safeguards

Most cyber tasks are handed to Opus 4.8, and biology research use needs the Life Sciences Verification Program; refusals return stop_reason: "refusal"

Model ID

claude-opus-5-5 (anthropic.claude-opus-5-5 on Bedrock; the same ID on Google Cloud and Microsoft Foundry)

Checked 2026-09-29 · Sources: Anthropic · Overview · Migration guide

Official benchmarks: Opus 5.5 vs Opus 5 vs Fable 5.1 vs Sonnet 5.5 vs GPT-6 Astra

From Anthropic’s launch page; these are vendor-reported results. Opus 5.5 ran at max effort (xhigh on Terminal-Bench) and GPT-6 Astra scores are OpenAI-reported. The Sonnet 5.5 column comes from the Sonnet 5.5 launch page; “—” means no score was published.

Anthropic notes that production safeguards were on during testing and cyber tasks fell back to Opus 4.8 when they triggered, which likely lowered Opus 5.5’s scores. In Artificial Analysis’ independent run, Opus 5.5 scored 59.6% on Terminal-Bench, level with GPT-6 Astra and 11 points above Opus 5, and 61.4% on Humanity’s Last Exam.

Which to use: Opus 5.5, Opus 5 or Sonnet 5.5

vs Opus 5

Cheaper and ahead on every published benchmark: Terminal-Bench 4.0 rose from 52.3% to 66.4% and GDPval-AA from 1708 to 1846. Anthropic says that at default effort it already beats Opus 5 at max on Terminal-Bench for about a fifth of the cost. Check the migration list below first: disabling thinking, manual budget_tokens and forced tool use all return 400, and the default effort dropped from high to medium.

vs Sonnet 5.5

Use Opus 5.5 for complex, open-ended work that needs sustained judgment and Sonnet 5.5 for well-scoped everyday coding and documents. Opus 5.5 lists at twice Sonnet 5.5’s price, but both read cache at $0.20. At max effort Artificial Analysis measured Sonnet 5.5 using about 60% more output tokens per task than Opus 5.5, so a task cost $7.60 on Sonnet 5.5 versus $5.98 on Opus 5.5. A common pattern is Opus 5.5 as the orchestrator with Sonnet 5.5 sub-agents.

Choosing effort

Anthropic recommends a fresh effort sweep on your own evals instead of carrying over Opus 5 settings, and a large max_tokens at higher levels, starting at 64K for xhigh and max. On Artificial Analysis’ Intelligence Index, medium scored 51 at $1.34 per task, high 54 at $1.82 and max 58 at $5.98; medium through max all sit on the cost/intelligence Pareto frontier. Community reports show max can exhaust the 128K output limit while still thinking, so it is rarely worth making the default.

Effort documentation

Third-party and community reviews

• Artificial Analysis: Intelligence Index 58 at max, #1 by several points; it leads six of the ten evaluations, and GDPval-AA is 111 Elo above Fable 5.1. At max it uses 1.6x Opus 5’s output tokens per task, but the price cut keeps a task at $5.98, level with Opus 5 (max) at $5.86. Every Wokey meter is 15% of the official rate, so the same task comes to about $0.90 at max and $0.20 at medium.
• Early-access customers (Anthropic launch page): Deloitte: caught 72% of known bugs in code review versus 56% for Opus 5 at high effort. Box: about a third of Opus 5’s tokens. Factory: the first model they would default to at medium effort. GitHub: among the fewest tokens and steps they measured. These are customer statements quoted by the vendor.
• Hacker News (1,804-point thread): Most of the thread debates the “pacing the frontier” wording, which many read as at odds with the release cadence. On the model itself, commenters describe it as a cost, efficiency and writing-style improvement rather than a capability jump, and welcome the price cut; one user saw max effort run out of the 128K token budget while still thinking, twice, with no answer.

Migrating from Opus 5: requests that return 400

These changes come from Anthropic’s migration guide. Check your requests and response parsing before switching the model to claude-opus-5-5.

• thinking: {"type": "disabled"} and {"type": "enabled", "budget_tokens": N} both return 400. Remove the thinking field or send {"type": "adaptive"}, and control depth with effort. Through Wokey the gateway strips these thinking fields before forwarding, so the request does not fail, but thinking still runs and bills as output tokens.
• tool_choice of any or tool returns 400, including on count_tokens. Use auto with strict: true or structured outputs; strict needs additionalProperties: false on every object in the schema.
• The default effort changed from high to medium, so a request without effort runs one level lower than on Opus 5. Set it explicitly.
• max_tokens covers thinking plus text. Re-size requests that used to disable thinking; start at 64K for xhigh or max.
• Responses can start with thinking blocks, so read text blocks by type instead of content[0].text. thinking.display defaults to "omitted" with empty text; set display: "summarized" to get readable summaries.
• Text between tool calls now comes back inside thinking blocks; to show progress, set display: "updates" (beta) or "summarized". Pass thinking blocks back unchanged in tool-use loops.
• Thinking blocks are bound to the model and the conversation: for accounts created on or after 2026-08-31, replaying them after editing history returns 400, and other models ignore them after a route change.
• The Claude API and Google Cloud reject computer_20251124; use computer_toolset_20260801 (Bedrock is unaffected). The minimum cacheable prompt drops to 512 tokens; non-default temperature, top_p or top_k and a trailing assistant prefill still return 400.

Request body after migration (/v1/messages)

{
  "model": "claude-opus-5-5",
  "max_tokens": 16000,
  "thinking": { "type": "adaptive", "display": "summarized" },
  "output_config": { "effort": "high" },
  "messages": [{ "role": "user", "content": "Review this diff for concurrency bugs" }]
}

Catalog pricing snapshot

Prices are per 1M tokens; request-time pricing controls settlement.

Official price source: https://www.anthropic.com/claude-opus-5-5

Model capabilities

Context

1,000,000 tokens

Max output

128,000 tokens

Streaming

Supported

Tools

Supported

Vision

Supported

Supported API forms

/v1/messages

Send a request

/v1/messages

curl https://api.wokey.ai/v1/messages \
  -H "x-api-key: $WOKEY_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{"model":"claude-opus-5-5","max_tokens":1024,"messages":[{"role":"user","content":"Hello"}]}'

Use Claude Opus 5.5 in your tools

The gateway is https://api.wokey.ai (OpenAI-compatible clients usually take https://api.wokey.ai/v1), and the model ID is claude-opus-5-5. Open the guide for the client you use.

• Use Claude Opus 5.5 in Claude Code: Anthropic Messages · base URL and API key setup
• Use Claude Opus 5.5 in Claude Desktop: Anthropic Messages · base URL and API key setup
• Claude API cost calculator: what Claude Opus 5.5 costs per month: Compare monthly cost on the Anthropic API, OpenRouter and Wokey for your usage and cache hit rate.
• OpenRouter alternative: Wokey vs OpenRouter: Compare token prices, payment fees, API forms, and upstream sources.
• Verifiable AI API: check a Claude Opus 5.5 response came from the official upstream: Which responses carry a tee.proof, and what a proof does and does not show.
• Verify a Claude Opus 5.5 response online: AI API relay verifier: Paste a response with its tee.proof and check it locally in your browser; nothing is uploaded.

Frequently asked questions

Related models and documentation

• Claude Opus 5
• Claude Opus 4.8
• Claude Opus 4.7
• Claude Opus 4.6
• API documentation
• All models and pricing
• Provider Node