What’s new in Claude Fable 5.2

Claude Fable 5.2 is built for demanding reasoning and long-horizon agentic work. For most workloads, start with the latest Claude Opus model (see Choosing the right model), and move to Claude Fable 5.2 when your evals on a Claude Opus model at higher effort still don’t meet your requirements.

Four changes from Claude Fable 5.1 can break your code:

• The prefix check applies to every account.
• Only Claude Fable 5.2 and Claude Mythos 5.2 read Claude Fable 5.2 thinking blocks.
• The earlier computer_20251124 computer use tool returns an error on every platform except Amazon Bedrock.
• With the advisor tool, a Claude Fable 5.2 executor needs a Claude Fable 5.2 or Claude Mythos 5.2 advisor.

Models

Claude Fable 5.2 and Claude Mythos 5.2 share these specs:

• Context window and output: a 1M token context window (default and maximum) and 128k max output tokens.
• Thinking: adaptive thinking is always on. Use the effort parameter to control thinking depth.
• Tokenizer: the tokenizer introduced with Claude Opus 4.7. See Token counting.

For all current models, see Models overview.

Breaking changes

Editing earlier turns invalidates thinking blocks on every account

The API’s prefix check (see Keeping the prefix unchanged) tests whether anything before a Claude Fable 5.2 or Claude Mythos 5.2 thinking block has changed since the model produced it: the system prompt, the tools, or an earlier message. On Claude Fable 5.2 and Claude Mythos 5.2, a request that sends back an invalid thinking block and leaves prefix_mismatch_behavior unset gets a 400 error, whatever the account’s age (see When prefix_mismatch_behavior is unset). On Claude Fable 5.1, such a request gets the error on accounts created on or after August 31, 2026. On the Claude API, Claude Mythos 5.1 already returns it on every account. So code on Claude Fable 5.1 that edits its history can run without errors on an older account before the switch and fail after it.

When the check fails, the request returns a 400 error whose message says The block is bound to a different conversation. To drop the affected blocks and continue instead, send the thinking-binding-controls-2026-08-01 beta header and set thinking.block_binding.prefix_mismatch_behavior to "drop_block". The response then lists each dropped block in a top-level input_transformations array. Treat "drop_block" as a stopgap while you fix the edit, because dropping blocks can raise output tokens.

Nothing changes for you if Claude Code, claude.ai, Claude Managed Agents, or the Claude Agent SDK builds your requests.

If your own code builds your requests, nothing changes for you either, as long as all of these are true:

• Your code keeps system and tools the same for the whole conversation.
• Your code only adds to messages. It never changes, reorders, or removes an earlier message.
• Your code sends each assistant turn back exactly as the API returned it: every block, in the same order.
• Your code does none of the things listed in What to check in your integration.

Otherwise, test your code before you switch:

• Code on Claude Fable 5.1: Follow the prefix check section of the migration guide.
• Code on Claude Mythos 5.1: Follow Migrating to Claude Mythos 5.2 from Claude Mythos 5.1.

Preserved thinking explains what counts as an edit. It also shows how to make each change without an edit.

Only Claude Fable 5.2 and Claude Mythos 5.2 read Claude Fable 5.2 thinking blocks

Every thinking block records which model produced it, and a model reads only its own blocks and some other models’ blocks. On the Claude API and Claude Models on Google Cloud (Google managed), Claude Fable 5.2 and Claude Mythos 5.2 read each other’s thinking blocks and those from Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5, Claude Fable 5.1, Claude Mythos 5.1, Claude Opus 5, Claude Fable 5, and Claude Mythos 5, and from earlier Opus, Sonnet, and Haiku models. No other model reads their blocks.

• Moving onto Claude Fable 5.2 or Claude Mythos 5.2 from one of those models keeps the conversation’s reasoning on the Claude API and Claude Models on Google Cloud (Google managed). So does moving between Claude Fable 5.2 and Claude Mythos 5.2.
• Moving from Claude Fable 5.2 or Claude Mythos 5.2 to any other model loses their reasoning for the turns that run there. That includes Claude Opus 5.5, Claude Fable 5.1, Claude Mythos 5.1, and Claude Fable 5.2’s fallback models (Claude Opus 4.8 and Claude Opus 5).

When a request carries a block the current model can’t read, the API drops the block before the model receives it. The request succeeds, and dropped blocks aren’t billed. With the thinking-binding-controls-2026-08-01 beta header, the response lists the drop in input_transformations. See Switching models mid-conversation.

Use the computer use toolset instead of computer_20251124

Claude Fable 5.1 accepts computer use both as the computer_toolset_20260801 toolset and, with the computer-use-2025-11-24 beta header, as the earlier computer_20251124 tool. Claude Fable 5.2 and Claude Mythos 5.2 support computer use through the toolset. A request to either model that declares a computer_20251124 tool returns a 400 invalid_request_error on every platform except Amazon Bedrock, where the earlier tool works as it does on Claude Fable 5.1.

To move an existing integration, follow Migrate from computer_20251124.

A Claude Fable 5.2 executor needs a Claude Fable 5.2 or Claude Mythos 5.2 advisor

With the advisor tool, which is in beta, the advisor must be at least as capable as the executor. A Claude Fable 5.2 or Claude Mythos 5.2 executor accepts only Claude Fable 5.2 or Claude Mythos 5.2 as its advisor. Any other advisor, including Claude Fable 5.1 and Claude Mythos 5.1, returns a 400 invalid_request_error.

If you use the advisor tool and move the executor to one of these models, set the model in the advisor tool definition to claude-fable-5-2 or claude-mythos-5-2 in the same change. Changing the advisor’s model is an edit to tools, so make the change in new conversations. A request that sends earlier thinking blocks back after the change fails the prefix check and returns a 400 error, unless you set thinking.block_binding.prefix_mismatch_behavior to "drop_block" (beta, thinking-binding-controls-2026-08-01 header). To keep your current executor and change only the advisor, set the advisor’s model to claude-fable-5-2: every executor model accepts it. A Claude Fable 5.2 advisor returns its advice encrypted, so if your code reads the advice text, see Result variants. See also Model compatibility.

Thinking and tool choice

Two Messages API rules apply to every Claude Fable 5.2 request. Neither is a change if you come from Claude Fable 5.1, Claude Mythos 5.1, or Claude Opus 5.5. From Claude Fable 5 or Claude Mythos 5, where adaptive thinking is already always on, only the forced tool use rule is a breaking change. From Claude Opus 5 or an earlier Claude Opus model, both are. The Claude Fable 5.2 and Claude Mythos 5.2 migration guide covers each path.

• Adaptive thinking is always on. thinking: {"type": "enabled"} with budget_tokens and thinking: {"type": "disabled"} both return a 400 error. Omit thinking or send {"type": "adaptive"}, and use effort to control thinking depth.
• Forced tool use isn’t supported. tool_choice set to {"type": "any"} or {"type": "tool", "name": "..."} returns a 400 invalid_request_error. For schema-valid JSON, keep tool_choice: {"type": "auto"} and set strict: true with strict tool use, or use JSON outputs. In an organization with customer-managed encryption keys, structured outputs aren’t available for Claude Fable 5.2 or Claude Mythos 5.2. That covers both strict tool use and JSON outputs. There, leave out strict and validate the tool input in your code.

Feature support

Claude Fable 5.2 supports these API features, among others:

• Effort: all five effort levels: low, medium, high, xhigh, and max. The API default is high. Start there, because most of the gain in capability is already present at high. Step up to xhigh for your hardest long-running coding and agentic work, where you’ve measured a gain. Reserve max for measuring the model’s ceiling and for the rare task where time and tokens don’t matter. See Recommended effort levels for Claude Fable 5.2.
• Compaction on demand (beta): with the compact-2026-09-04 beta header, you choose when to replace older turns with a summary. See Compaction on demand.
• Mid-conversation tool changes (beta): you add or remove tools in a role: "system" message instead of editing tools. On the Claude API, Claude Platform on AWS, and Claude Models on Google Cloud (Anthropic managed), send the inline-tools-2026-09-15 beta header. The mid-conversation-tool-changes-2026-07-01 header works for changes that name a tool by reference, on the platforms that Mid-conversation system messages and tool changes lists. See Mid-conversation tool changes.
• Computer use: through the computer_toolset_20260801 toolset. The earlier computer_20251124 tool returns a 400 error on every platform except Amazon Bedrock, where both models also accept the earlier tool (see the computer use breaking change). See Computer use tool.

Compaction on demand and mid-conversation tool changes help you keep the history append-only, which the prefix check requires. See each feature’s page for model and platform availability.

Fast mode is available for Claude Fable 5.2 in Messages API requests on the Claude API.

Fast mode is also available for Claude Fable 5.2 on Claude Models on Google Cloud (Anthropic managed).

Fast mode is also available for Claude Fable 5.2 on Claude Models on Google Cloud (Google managed).

Fast mode is also available for Claude Fable 5.2 on Claude in Microsoft Foundry for deployments hosted on Anthropic, as a separate model, claude-fable-5-2-fast.

Fast mode is also available for Claude Mythos 5.2 in Messages API requests on the Claude API. Availability of fast mode lists every platform where Claude Mythos 5.2 has fast mode.

Thinking blocks stay with the account that produced them

Thinking blocks that Claude Fable 5.2 and Claude Mythos 5.2 produce work only in the account that produced them, or in an account linked to it. When another account sends one of these blocks, the API drops the block before the model receives it, and the request succeeds. On the Claude API and Claude Models on Google Cloud (Google managed), with the thinking-binding-controls-2026-08-01 beta header, the response lists each dropped block in input_transformations with reason: "organization_binding_mismatch". See Thinking blocks stay with the account that produced them in Preserved thinking.

Behavior differences

A few Claude Fable 5.2 behaviors show up without any code change. Check them when you carry prompts over. Prompting Claude Fable 5.2 has guidance for each:

• Prompt lines that ask for less thinking don’t stop the model from thinking at medium effort and above. Claude Fable 5.1 usually follows a line such as “answer directly without deliberating.” In Anthropic’s testing, such a line didn’t stop Claude Fable 5.2 from thinking at medium and above. At high and above, the model thinks before almost every reply. If easy requests take too long or cost too much, use a lower effort level for those turns. See Calibrate effort.
• A turn can end before the work is done. On long, multistep tasks, the model sometimes ends a turn with a list of what remains, an offer to continue, or a stop at an ordinary obstacle. This occurs more often at low and medium effort. See Finish the whole task.
• What the prompt says about code ownership matters. In Anthropic’s testing on small coding tasks, when a prompt said nothing about who owns the code, the model tended to add to the existing structure and to report nearby problems instead of fixing them. See Say who owns the code.
• Long runs of tool calls can pass with little user-facing text. The model might also make its first tool call without saying what it’s about to do. See Keep users informed during long turns.
• How much the model delegates to subagents depends on your harness and tasks. In an agentic coding product whose prompts invite delegation, the model delegates readily. Subagents can account for a large part of what a session costs. See Set expectations for subagents.
• The model follows style rules closely. A hard limit on sentence length can make the model’s prose choppy. See Review instructions written for earlier models.

Refusals, fallback, and billing

Claude Fable 5.2 runs safety classifiers, and everything in Refusals and fallback applies.

• Refusals: a declined request returns HTTP 200 with stop_reason: "refusal" and a stop_details object naming the policy area that fired.
• Fallback: retry a refused request on another model with server-side fallback, the SDK middleware, or your own retry. Server-side fallback is in beta on the Claude API and Claude Models on Google Cloud (Anthropic managed). With it, fallbacks: "default" retries a declined request on the model Anthropic recommends for that category. For categories with no recommended fallback, such as reasoning_extraction, the refusal stands. The permitted fallback targets for Claude Fable 5.2 are Claude Opus 4.8 and Claude Opus 5.
• Billing: whether a refusal that arrives before any output is billed depends on its refusal category, and it counts against your rate limits either way. See How refusals are billed. Fallback credit refunds the prompt-cache cost of switching models.

Pricing

Claude Fable 5.2 and Claude Mythos 5.2 share these prices, in USD:

Input and output cost less than on Claude Fable 5.1 and Claude Mythos 5.1, which cost $10 USD per million input tokens and $50 USD per million output tokens. Cache reads (hits and refreshes) cost 0.025 times the base input price, the same multiplier as on those models. Batch processing is $4 USD per million input tokens and $20 USD per million output tokens. Fast mode for Claude Fable 5.2 costs $16 USD per million input tokens and $80 USD per million output tokens. See Fast mode pricing. Fast mode for Claude Mythos 5.2 costs $16 USD per million input tokens and $80 USD per million output tokens. See Pricing for data residency and tool pricing.

Availability

Claude Fable 5.2 is available on:

• Claude API: all customers, as claude-fable-5-2
• AWS: Claude in Amazon Bedrock, as anthropic.claude-fable-5-2, and Claude Platform on AWS, as claude-fable-5-2
• Google Cloud: Claude Models on Google Cloud (Anthropic managed), as claude-fable-5-2, and Claude Models on Google Cloud (Google managed), as claude-fable-5-2
• Microsoft Foundry: Claude in Microsoft Foundry, as claude-fable-5-2, on Anthropic infrastructure (see Hosting options)

Claude Mythos 5.2 is available only to organizations verified through Anthropic’s verification programs, such as the Cyber Verification Program. It’s available on:

• Claude API: as claude-mythos-5-2
• AWS: Claude in Amazon Bedrock, as anthropic.claude-mythos-5-2
• Google Cloud: Claude Models on Google Cloud (Anthropic managed), as claude-mythos-5-2, and Claude Models on Google Cloud (Google managed), as claude-mythos-5-2
• Microsoft Foundry: Claude in Microsoft Foundry, as claude-mythos-5-2, on Anthropic infrastructure

To request access, apply to the program that covers your use case, or contact your account team.

Claude Fable 5.2 and Claude Mythos 5.2 require 30-day data retention and aren’t available under zero data retention unless expressly authorized by Anthropic. Both are Covered Models. See Model-specific data retention requirements.

Migrate from Claude Fable 5.1

To migrate from Claude Fable 5.1, update your model ID:

model = "claude-fable-5-1"  # Before
model = "claude-fable-5-2"  # After

Then review these items:

1. Check whether your code edits history. If your code builds the messages array itself, run the test in the migration guide’s prefix check section, even if Claude Fable 5.1 has never returned the error for your account.
2. Check model switches. If a router or fallback moves conversations between models, expect every model except Claude Mythos 5.2 to drop Claude Fable 5.2’s thinking blocks, Claude Fable 5.1 and Claude Opus 5.5 included. See Only Claude Fable 5.2 and Claude Mythos 5.2 read Claude Fable 5.2 thinking blocks.
3. Move computer use to the toolset. If you use the computer_20251124 tool anywhere except Amazon Bedrock, Claude Fable 5.2 returns a 400 error for that tool. Declare computer_toolset_20260801 instead. For the change itself, see Use the computer use toolset instead of computer_20251124.
4. Update the advisor. If you use the advisor tool with Claude Fable 5.2 as the executor, set the advisor’s model to claude-fable-5-2 or claude-mythos-5-2. See A Claude Fable 5.2 executor needs a Claude Fable 5.2 or Claude Mythos 5.2 advisor.
5. Re-run your evals. Include an effort sweep that starts at the default, high. See Recommended effort levels for Claude Fable 5.2.

See the Claude Fable 5.2 and Claude Mythos 5.2 migration guide for step-by-step instructions, including from Claude Mythos 5.1 and from Claude Opus models. For model-specific prompting patterns, see Prompting Claude Fable 5.2.

Next steps