How do I migrate from Claude Fable 5.1 to Fable 5.2?

Change the model ID from claude-fable-5-1 to claude-fable-5-2 (anthropic.claude-fable-5-2 on Amazon Bedrock), then check four breaking changes Anthropic lists for the November 9, 2026 release: edited conversation history returns a 400 error on every account, Fable 5.2 thinking blocks can only be read by Fable 5.2 and Mythos 5.2, the computer_20251124 tool is rejected outside Bedrock, and an advisor paired with a Fable 5.2 executor must itself be Fable 5.2 or Mythos 5.2.

Why do I get 'The block is bound to a different conversation' on Fable 5.2?

Your code changed something before a thinking block that Fable 5.2 produced: the system prompt, the tools list or an earlier message. Fable 5.2 enforces this prefix check on every account; Fable 5.1 enforced it only on accounts created on or after August 31, 2026. Fix the code so it keeps system and tools fixed, only appends messages and sends assistant turns back exactly as returned. As a stopgap, send the thinking-binding-controls-2026-08-01 beta header with prefix_mismatch_behavior set to drop_block.

Is Claude Fable 5.2 cheaper than Fable 5.1?

Yes. As of November 10, 2026 Fable 5.2 costs $8 per million input tokens and $40 per million output tokens, against $10/$50 for Fable 5.1. Cache reads fall from $0.25 to $0.20, and Batch is $4/$20. Fable 5.2 also uses fewer tokens per task at the default medium effort, according to Anthropic, so the saving per task can exceed 20%.

Do I need to change anything if I use Claude Code?

No code changes are needed if Claude Code, claude.ai, Claude Managed Agents or the Claude Agent SDK builds your requests. Anthropic says the prefix-check change only affects code that assembles its own Messages API requests and edits, reorders or removes earlier turns.

Quick Answer

How to Migrate to Claude Fable 5.2: Breaking Changes Guide

Published: November 10, 2026

The short answer

Swap the model ID to claude-fable-5-2, then test four things: history edits, model switching, computer use and advisor pairs. Anthropic released Claude Fable 5.2 and Mythos 5.2 on November 9, 2026 at $8/$40 per million tokens, 20% below Fable 5.1. If Claude Code, claude.ai, Managed Agents or the Claude Agent SDK builds your requests, the ID change is all you need.

Step 1: change the model ID

Platform	Fable 5.2 ID	Mythos 5.2 ID (verified orgs)
Claude API	`claude-fable-5-2`	`claude-mythos-5-2`
Amazon Bedrock	`anthropic.claude-fable-5-2`	`anthropic.claude-mythos-5-2`
Claude Platform on AWS	`claude-fable-5-2`	—
Google Cloud (Anthropic or Google managed)	`claude-fable-5-2`	`claude-mythos-5-2`
Microsoft Foundry (Anthropic hosted)	`claude-fable-5-2`	`claude-mythos-5-2`

Specs stay familiar: 1M-token context window, 128K max output, adaptive thinking always on (control depth with the effort parameter) and the tokenizer introduced with Opus 4.7.

Step 2: fix history edits (the 400 error)

The API checks that nothing before a Fable 5.2 thinking block changed since the model wrote it: the system prompt, the tools list or an earlier message. On Fable 5.2 a mismatch returns 400 — “The block is bound to a different conversation” on every account. Fable 5.1 only enforced it on accounts created on or after August 31, 2026, so code that worked on an older account can break after the switch.

You are safe if your code:

keeps system and tools identical for the whole conversation;
only appends to messages, never edits, reorders or deletes;
sends each assistant turn back exactly as returned, every block in order.

Common offenders are context-trimming middleware, prompt templates that inject a timestamp into the system prompt, and tool lists built dynamically per turn. While you fix them, send the thinking-binding-controls-2026-08-01 beta header and set thinking.block_binding.prefix_mismatch_behavior to "drop_block". Dropped blocks are listed in input_transformations. Anthropic warns that dropping blocks can raise output tokens, so treat it as temporary.

Step 3: plan model switching

Fable 5.2 and Mythos 5.2 can read thinking blocks from older Claude models (Opus 5.5, Sonnet 5.5, Haiku 5.5, Fable 5.1 and earlier) on the Claude API and Google Cloud. No other model can read Fable 5.2’s blocks. If your router falls back from Fable 5.2 to Opus 5.5, the fallback turn runs without the earlier reasoning. The request still succeeds and unreadable blocks are not billed, but quality can drop. Route a conversation to one model where you can.

Step 4: move computer use to the toolset

Fable 5.2 supports computer use only through computer_toolset_20260801. Declaring the older computer_20251124 tool returns a 400 invalid_request_error on every platform except Amazon Bedrock. Update the tool definition before switching any computer-use agent.

Step 5: check advisor pairs

With the beta advisor tool, a Fable 5.2 executor needs a Fable 5.2 or Mythos 5.2 advisor. The more common pattern runs the other way: Haiku, Sonnet or Opus executors consult Fable 5.2 as the advisor, which Anthropic suggests as a way to buy Fable-level judgment at a lower price.

Step 6: re-check cost and compliance

	Fable 5.1	Fable 5.2
Input / output per MTok	$10 / $50	$8 / $40
Cache read	$0.25	$0.20
5m / 1h cache write	$12.50 / $20	$10 / $16
Batch	$5 / $25	$4 / $20
Fast mode	—	$16 / $80

Both Fable 5.2 and Mythos 5.2 require 30-day data retention and are not available under zero data retention unless Anthropic expressly authorizes it. If your contract depends on ZDR, check before you switch. Compare vendors on our current API prices table, and see Fable 5.2 vs Mythos 5.2 vs Opus 5.5 if you are not sure Fable is the right tier. Our earlier Fable 5.1 migration guide covers the jump from Fable 5.

Migration checklist

Model ID updated per platform
Replay a long multi-turn conversation from your logs and confirm no 400s
Router does not bounce mid-conversation between Fable 5.2 and other models
Computer-use agents declare computer_toolset_20260801
Evals rerun at medium and high effort; token usage compared with Fable 5.1
Data-retention terms confirmed

Last verified: November 10, 2026.