From Claude Opus 5 to Opus 5.5

1. Breaks

   You can no longer switch thinking offi

   Telling the model not to think now makes the request fail. Use a lower effort settingi instead.

   Affects you if you turned thinking off to save time or money.

   Show the fix

   Opus 5 accepted thinking: {"type": "disabled"} at effort high or below. Opus 5.5 rejects it at every level.

   The error you'll see"thinking.type.disabled" is not supported for this model. Use "thinking.type.adaptive" and "output_config.effort" to control thinking behavior.

   1. Delete the thinking field, or send {"type": "adaptive"}, which means the same thing.
   2. Where you turned thinking off for speed, set effort to low instead and measure.
   3. Read the reply by block type. A response can now start with a thinking blocki before the text.
   4. Remove prompt lines that asked the model to "show its reasoning" in the answer. Opus 5.5 can refuse those as reasoning extractioni.

   client.messages.create(
       model="claude-opus-5",
       max_tokens=2000,
       thinking={"type": "disabled"},
       messages=[{"role": "user", "content": "Summarise this invoice"}],
   )

   client.messages.create(
       model="claude-opus-5-5",
       max_tokens=2000,
       output_config={"effort": "low"},  # effort is the only dial now
       messages=[{"role": "user", "content": "Summarise this invoice"}],
   )

   Source: What's new in Claude Opus 5.5 ↗

2. Breaks

   You can no longer force a tool call with tool_choicei

   Forcing the model to use one particular tooli now makes the request fail. You ask for it in the prompt instead.

   Affects you if you forced a tool call to get clean JSON back, or to make sure a step always runs.

   Show the fix

   Opus 5 accepted tool_choicei set to any or tool. Opus 5.5 rejects both, on the Messages API, Batches and token counting.

   The error you'll seetool_choice: type "tool" and "any" are not supported for this model.

   1. If the forced call only existed to get JSON, move the schema to structured outputsi.
   2. Otherwise use tool_choice: {"type": "auto"}, mark the tool strict: true (strict tool usei), and say in the prompt when to use it.
   3. Check the reply for a tool_use block and retry if it is missing. auto does not guarantee a call.

   tool_choice={"type": "tool", "name": "extract_invoice"}

   tools=[{**extract_invoice, "strict": True}],
   tool_choice={"type": "auto"},
   # and in the prompt: "Record the invoice with the extract_invoice tool."
   # then: if no tool_use block came back, retry once

   Source: What's new in Claude Opus 5.5 ↗

3. Breaks

   Editing earlier messages can make requests faili

   If your app rewrites or trims earlier messagesi in a conversation, the next request can fail. Add new messages instead of editing old ones.

   Affects you if your app edits or trims earlier messages, and your Anthropic account was created on or after 31 Aug 2026.

   Show the fix

   Each Opus 5.5 thinking blocki is tied to everything before it: the system prompt, the tool list and every earlier message. Change any of those, send the old block back, and the request fails. Opus 5 never checked this.

   The error you'll seeInvalid `signature` in `thinking` block. The block is bound to a different conversation.

   1. Make the conversation append-only. Add new instructions as a mid-conversation system messagei instead of editing the system prompt.
   2. Declare every tool in the first request. Add or remove tools with tool_addition / tool_removal blocks (beta).
   3. Trim long conversations with server-side compactioni, which does not count as an edit.
   4. Older accounts are not blocked, but a request opts in when it sets thinking.block_binding.prefix_mismatch_behavior (with the thinking-binding-controls-2026-08-01 beta header). Set it in your test pipeline to catch edits early.

   # request 1
   system = "You are a support agent."
   # request 2, same conversation
   system = "You are a support agent. The user is on the Pro plan."

   messages.append({
       "role": "system",
       "content": "The user is on the Pro plan.",
   })

   Source: Preserved thinking ↗

4. Breaks

   The older computer-use tool is rejectedi

   Agentsi that operate a computeri (click, type, take screenshots) need the newer tool setup, or the request fails.

   Affects you if you built an agent that controls a computer through screenshots and clicks.

   Show the fix

   Opus 5 accepted both the toolset and the older computer_20251124 tool. On the Claude API and Google Cloud, Opus 5.5 accepts only the toolset. On Amazon Bedrock the older tool still works.

   The error you'll see'claude-opus-5-5' does not support tool types: computer_20251124.

   1. Replace the tool entry with {"type": "computer_toolset_20260801"}. No name, no screen size, no beta header.
   2. Update the agent loop. The action now arrives as the tool block's name (screenshot, left_click), several can arrive in one turn, and every result must include "toolset_name": "computer".
   3. Make the change on your current model first and test it there.

   tools=[{"type": "computer_20251124", "name": "computer",
           "display_width_px": 1024, "display_height_px": 768}]

   tools=[{"type": "computer_toolset_20260801"}]

   Source: Computer use: migrate from computer_20251124 ↗

5. Changes quietly

   The default effort drops from high to mediumi

   If you never picked an effort leveli, Opus 5.5 works one notch lighter than your old model did.

   Show the fix

   A request that does not set efforti ran at high on Opus 5. On Opus 5.5 it runs at medium.

   1. Set output_config.effort on every request, so the level lives in your code and not in a default that can change.
   2. Start at medium, then test low and high on your own tasks. Keep the cheapest level that still passes.
   3. Anthropic documents the medium default for the Claude API. It does not state the default on the cloud platformsi, which is one more reason to set it yourself.

   Source: Effort: recommended levels for Claude Opus 5.5 ↗

6. Changes quietly

   The same effort name buys a different amount of worki

   The effort level you used before may be more than you need now. Test a lower one before copying it over.

   Affects you if you copy your Opus 5 effort setting across unchanged. Expect slower, longer replies.

   Show the fix

   At the same level, Opus 5.5 thinks more per turn than Opus 5, most of all at xhigh and max. In Anthropic's testing, Opus 5.5 at medium matched or beat Opus 5 at high.

   1. Run an effort sweepi: the same 10 to 20 real tasks at low, medium and high.
   2. To get less thinking, lower effort first. Anthropic says this works more reliably than a "think less" instruction.
   3. Reserve xhigh and max for work where you measured a gain.

   Source: Prompting Claude Opus 5.5: calibrate effort ↗

7. Changes quietly

   Switching models mid-conversation can lose contexti

   Move a conversation from one modeli to another halfway through, and the new model may start without the earlier model's reasoning.

   Show the fix

   Moving from Opus 5 onto Opus 5.5 keeps the earlier reasoning. A fallback from Opus 5.5 to Opus 5 or Opus 4.8 runs without Opus 5.5's reasoning.

   1. Keep passing thinking blocks back unchanged. The API drops Opus 5.5 blocks the target cannot read, unbilled, so stripping them yourself saves nothing.
   2. Expect the first turn after a switch to cost more, because the new model re-plans.
   3. Use the reasoning handoff tool further down this page to check a specific switch.

   Source: Preserved thinking: switching models ↗

8. Changes quietly

   Your AI agent's progress updates go quieti

   The short progress notes your app showed while an ai agenti worked are now hidden unless you switch progress updatesi on.

   Affects you if your interface shows those notes to users as live progress.

   Show the fix

   The notes between tool calls came back as text on Opus 5. On Opus 5.5 they come back inside thinking blocksi that are empty by default. Nothing errors. The screen just stops updating.

   1. Set thinking: {"type": "adaptive", "display": "updates"} with the thinking-display-updates-2026-08-18 beta headeri.
   2. Render every non-empty thinking block as a status line, ahead of the tool call it introduces.
   3. For updates at set points, say so in the system prompt, or send the check-in.txt prompt further down this page.

   client.beta.messages.create(
       model="claude-opus-5-5",
       betas=["thinking-display-updates-2026-08-18"],
       thinking={"type": "adaptive", "display": "updates"},
       ...
   )

   Source: Opus 5.5 migration guide: text between tool calls ↗

9. Changes quietly

   Some requests get refused or answered by an older modeli

   A small set of sensitive requests (security testing, some biology, some AI-chip work) is refused or handed to an older modeli.

   Affects you if your product touches life sciences, ML kernels or accelerators, or security.

   Show the fix

   On Opus 5, cyber was the only category that fell back to another model (Opus 4.8). Opus 5.5 adds bio and frontier LLM worki, both falling back to Opus 5, and a reasoning_extraction refusal category on the API that is never retried.

   1. Check stop_reason before reading content. Code that reads content[0] breaks on an empty refusal.
   2. On the API, fallbacksi are off until you opt in. Send fallbacks: "default" with the server-side-fallback-2026-07-01 beta headeri.
   3. On Amazon Bedrock, Google Cloud and Microsoft Foundry there is no server-side fallback. Use the SDK's fallback middleware, which runs in your own code on any platform.
   4. Log which model answered (response.model). Your quality and cost numbers include the fallback model on those turns.
   5. Reasoning-extraction refusals are never retried on another model. Remove prompts that ask for the model's raw reasoning.

   response = client.beta.messages.create(
       model="claude-opus-5-5",
       betas=["server-side-fallback-2026-07-01"],
       fallbacks="default",
       ...
   )
   if response.stop_reason == "refusal":
       ...  # the whole chain declined
   print(response.model)  # which model actually answered

   Source: Why Claude switched models (Help Center) ↗

10. Good news

    Every token costs lessi

    Input drops from $5 to $4 per million tokensi and output from $25 to $20. Cache readsi drop from $0.50 to $0.20, the biggest cut.

    What to do with it

    1. Re-measure cost per finished task, not per token. Opus 5.5 thinks more per effort level but, in Anthropic's testing, finishes in fewer steps.
    2. Keep the prompt cache warm. Cached input costs $0.20 per million tokens against $4 for fresh input, 20 times as much.

    Source: Pricing ↗

11. Good news

    Fast mode costs less, or exists at alli

    Fast modei costs $10 / $50 per million tokens on Opus 5 and $8 / $40 on Opus 5.5. On the Claude API and in Claude Code, not on the cloud platformsi.

    Source: Fast mode ↗

12. Good news

    Replies come back sooneri

    Anthropic says Opus 5.5 writes output more than 30% faster than Opus 5 and tends to finish the same task with fewer tokens.

    Source: Prompting Claude Opus 5.5 ↗

13. Good news

    It knows about more recent eventsi

    The reliable knowledge cutoffi moves from May 2026 to June 2026.

    Source: Models overview ↗

14. Good to know

    You draw from a different rate-limit pooli

    Opus 5 and Opus 5.5 each have their own rate limiti of the same size. Running both during a migration uses two pools.

    Source: Rate limits ↗

That's every change. You're on Opus 5.5.