Skip to content
Anish Gandhi Claude Opus 5.5 Connect on LinkedIn

← All guides · checked against Anthropic's docs on

Opus 5.5 on the API: pick your model

For products that call Claude from their own code, directly or through a cloud platform. Pick the model you use today to see only what changes for you, with the fix one tap away.

Using Claude yourself instead? Chat · Cowork · Claude Code

New to this? Start here

Opus 5.5 is the newest Claude . If your app uses the , a few settings it sends may now fail or behave differently. This page lists each one, with the fix.

Opus 5.5 at a glance

Four facts to hold on to, whichever model you come from.

  • $4 / $20per input / output tokensOpus 4.8 and 5: $5 / $25 · Fable 5.1: $10 / $50
  • mediumdefault effort when you send nonehigh on all three older models
  • Always onthinking, with no way to switch it offOpus 4.8: off unless asked · Opus 5: on, can be switched off
  • June 2026reliable knowledge cutoffOpus 4.8: Jan 2026 · Opus 5: May 2026

What it will cost you

Type in what you pay today and pick your model.

The same work on Opus 5.5: about $800 a month, $200 less.

At Anthropic's list prices, for the same number of tokens. Each model uses a different number of tokens for the same task, so check your real bill after a week.

Work it out by token type
Fable 5.1 $415
Opus 5 $230
Opus 4.8 $230
Opus 5.5 $172

Same token counts on every model. Real counts differ: see the note below.

Token counts will not stay the same. Opus 5.5 thinks more per effort level than Opus 5 but, in Anthropic's testing, finishes tasks in fewer steps and fewer tokens. Anthropic's launch post puts typical workloads at about 40% cheaper than Opus 5. Measure cost per finished task on your own traffic.

Unit prices per million tokens
PriceFable 5.1Opus 5Opus 4.8Opus 5.5
Input $10$5.00$5.00$4.00
Output $50$25$25$20
Cache read $0.25$0.50$0.50$0.20
5-min cache write $13$6.25$6.25$5.00
Batch output $25$13$13$10
Fast mode (in / out) Not offered$10 / $50$10 / $50$8.00 / $40

Questions developers ask

How much does Claude Opus 5.5 cost?

$4 per million input tokens and $20 per million output tokens. Opus 5 and Opus 4.8 cost $5 and $25. Fable 5.1 costs $10 and $50. Cache reads on Opus 5.5 cost $0.20 per million tokens.

Why does tool_choice "any" fail on Opus 5.5?

Opus 5.5 does not support forced tool use. tool_choice set to "any" or "tool" returns a 400. Use "auto", mark the tool strict: true, and say in the prompt when to call it.

Does Opus 5.5 keep the reasoning from an Opus 5 conversation?

Yes. Opus 5.5 reads thinking blocks from Opus 5 and earlier Opus, Sonnet and Haiku models. It cannot read Fable blocks, and on the Claude API only Fable 5.1 and its Mythos 5.1 sibling can read its own.

Does Opus 5.5 support computer use?

Yes, through the computer_toolset_20260801 toolset. On the Claude API and Google Cloud it rejects the older computer_20251124 tool. On Amazon Bedrock the older tool still works.

What happens when Opus 5.5 refuses a request?

The API returns a normal response with stop_reason "refusal". If you opt in to fallbacks, cyber refusals are retried on Opus 4.8, and bio and frontier LLM refusals on Opus 5. Reasoning-extraction refusals are not retried.

For whoever wires it up

The technical details

Everything above is enough to decide and to brief your developer. This part is the precise version: settings, model switches, refusal routing and a full spec sheet.

Thinking and effort on each model

Thinking is locked on. Effort is the only control, and its default is now medium.

Can thinking be switched off?

  • Opus 4.8 Off unless you askOmit the field and it does not think
  • Opus 5 On by defaultCan be switched off at effort high or below
  • Fable 5.1 Always onSending "disabled" returns a 400
  • Opus 5.5 Always onSending "disabled" returns a 400

How hard does it work by default?

Opus 4.8 low medium high default xhigh start here max
Opus 5 low medium high default xhigh max
Fable 5.1 low medium high default xhigh max
Opus 5.5 low medium default high xhigh max

Level names do not mean the same work across models. In Anthropic's testing, Opus 5.5 at medium matched or beat Opus 5 at high on coding and knowledge work, and low came close on several coding tests.

Switching models mid-conversation

Each model keeps private notes on why it did what it did. When a conversation moves to another model, those notes are either passed on or thrown away.

Moving a conversation from to

Opus 5.5 gets Opus 5's notes, so it knows why earlier decisions were made.

See every combination
Rows: the model the conversation started on. Columns: the model it moves to.
Fromto Opus 4.8to Opus 5to Fable 5.1to Opus 5.5
from Opus 4.8 Same model Not documented Notes passed on Notes passed on
from Opus 5 Not documented Same model Notes passed on Notes passed on
from Fable 5.1 Notes thrown away Notes thrown away Same model Notes thrown away
from Opus 5.5 Notes thrown away Notes thrown away Passed on (Claude API) Same model

Where a flagged request goes

Opus 5.5 checks every request, and everything the model reads, against four named categories.

Opus 5.5 screens every request . Almost everything goes straight through. A few sensitive topics are answered by an older model instead, and one kind is refused.

  • Everyday work Writing and reviewing code, finding bugs in your own code, everyday health and biology questions Opus 5.5 answers
  • Hacking and security testing Writing exploits, scanning compiled programs for weaknesses, penetration testing Opus 4.8 answers instead
  • Risky biology Dual-use virology, toxicology or molecular design Opus 5 answers instead
  • Building other AI models A small set of tasks, such as low-level code for certain AI chips Opus 5 answers instead
  • Asking for its private reasoning Asking it to paste its internal reasoning word for word Refused, no retry

In the Claude app, Cowork and Claude Code the switch happens for you, with a notice saying which model answered. One setting turns it into a pause instead.

In your own app nothing switches unless your developer turns it on. Until then the request comes back refused.

For developers

A refusal is a normal HTTP 200 with stop_reason: "refusal" . Opt in with fallbacks: "default" and the server-side-fallback-2026-07-01 beta header; response.model tells you which model answered. Opus 5.5 is not yet in Anthropic's Cyber Verification Program. Life-sciences teams can apply to the Life Sciences Verification Program.

On Bedrock, Google Cloud or Foundry

The same model, but not every feature arrives everywhere.

Opus 5.5 features by platform
FeatureClaude APIAmazon BedrockGoogle CloudMicrosoft Foundry
Model ID claude-opus-5-5 anthropic.claude-opus-5-5 claude-opus-5-5 Your deployment name
Refused requests handed to another model on the server Beta, opt in Use SDK middleware Use SDK middleware Use SDK middleware
Old computer-use tool (computer_20251124) Rejected Still works Rejected Earlier versions only
Fast mode Preview No No No
Batch (half price) Yes No No No
Web search Yes No Yes Basic only on Azure hosting
Web fetch Yes No No Basic only on Azure hosting
Beta header anthropic-beta Passed differently Passed differently Passed differently
Medium effort default Documented Not stated Not stated Not stated
Prompt caching and 1M context Yes Yes Yes Yes

The breaking changes on this page (thinking locked on, no forced tool calls, the edited-history check) apply on every platform. Claude Platform on AWS follows the Claude API, except that it has no fast mode and only the earlier computer-use tools. Anthropic's pages disagree on whether Bedrock supports structured outputs, so test it before relying on it.

Source: Features overview and each platform's page

All four models on one sheet

Read across a row to see how one setting moved from model to model.

Property Opus 4.8 Opus 5 Fable 5.1 Opus 5.5 new
Basics
API model ID claude-opus-4-8claude-opus-5claude-fable-5-1claude-opus-5-5
Released 28 May 202624 Jul 20261 Sep 202622 Sep 2026
Reliable knowledge cutoff Jan 2026May 2026Jun 2026Jun 2026
Latency (Anthropic rating) Not rated in the current lineupNot rated in the current lineupSlowerModerate
Retirement Not before 28 May 2027Not before 24 Jul 2027Not before 1 Sep 2027Not before 22 Sep 2027
Price per million tokens
Input / output $5 / $25$5 / $25$10 / $50$4 / $20
Cache read $0.50$0.50$0.25$0.20
Cache write, 5 min / 1 hr $6.25 / $10$6.25 / $10$12.50 / $20$5 / $8
Batch input / output $2.50 / $12.50$2.50 / $12.50$5 / $25$2 / $10
Fast mode input / output $10 / $50$10 / $50Not offered$8 / $40
Thinking and effort
No thinking field sent Does not thinkThinksThinksThinks
Can thinking be switched off YesAt effort high or belowNo (400 error)No (400 error)
Default effort highhighhighmedium
Anthropic's suggested start xhigh for coding and agentshigh, then sweep downhighmedium, then test low and high
API behaviour
Force a tool call (any / tool) AllowedAllowedNo (400 error)No (400 error)
Edited history checked NoNoYes, for accounts from 31 Aug 2026Yes, for accounts from 31 Aug 2026
Notes between tool calls text blockstext blocksthinking blocksthinking blocks
Minimum cacheable prompt 1,024 tokens512 tokens512 tokens512 tokens
Safety and account
Safety classifiers None of its own. It is the model other Opus models fall back toCyber, falling back to Opus 4.8Cyber, bio, frontier LLM work and reasoning extractionCyber (to Opus 4.8), bio and frontier LLM work (to Opus 5), reasoning extraction blocked
Data retention Zero data retention availableZero data retention available30-day retention required. Zero data retention only if Anthropic authorises itZero data retention available
Priority Tier YesNoNoNo
Rate-limit pool Shared with Opus 4.7, 4.6 and 4.5Its own poolShared with Fable 5Its own pool
Start tier input tokens / min 2,000,0002,000,000500,0002,000,000
Same on all four
Context window / max output 1M / 128K1M / 128K1M / 128K1M / 128K
Tokenizer SameSameSameSame

Sources

Every fact on this page comes from Anthropic's own documentation, checked on 23 Sep 2026, unless it is labelled as a tester's report. Numbers from testers come from one small team with early access. Treat them as early impressions.

Anthropic's documentation

Early testers

Press