Claude Fable 5.2

Claude Fable 5.2 is Anthropic’s November 9 model for demanding reasoning and long-horizon agents. Use claude-fable-5-2 on the Claude API or the separately documented provider ID on your chosen host.

Claude Fable 5.2 guide
Upgrade decisions, Fable/Mythos boundaries, migration risks and publisher demonstrations.
Claude Fable 5.1
Preserve the predecessor’s separate ID, rates and contract.
Claude Opus 5.5
Compare the lower-priced daily-driver model.

Language ModelCoding AssistantAgent WorkflowsAdaptive ThinkingLong ContextAnthropic

Hosted Claude API and documented partner platforms; no self-hosted checkpoint verified here.

Standard Claude API USD/M tokens: $8 input, $40 output; $0.20 cache read, $10 5m/$16 1h cache write. Fast mode $16/$80 on supported platforms.

Commercial

🚀Function Overview

A 1M-context, 128K-output model with always-on adaptive thinking. Standard Claude API prices are $8 input and $40 output per million tokens; account and platform restrictions still apply.

Key Features

Claude API: claude-fable-5-2; Amazon Bedrock: anthropic.claude-fable-5-2; OpenRouter: anthropic/claude-fable-5.2
Text and image input; text output, including supported tool calls
1,000,000-token context window and 128,000 maximum output tokens
Adaptive thinking always on; API default high, with low, medium, high, xhigh and max supported
Standard USD/M tokens: input $8, output $40, cache read $0.20, 5m cache write $10, 1h cache write $16
Optional fast mode: $16 input and $80 output per million tokens on supported surfaces
30-day retention required unless Anthropic expressly authorizes an exception; check platform-specific terms
Preserved thinking prefix binding applies to every account; only Fable/Mythos 5.2 read their own 5.2 thinking blocks
Old computer_20251124 is unsupported except on Bedrock; use computer_toolset_20260801
OpenRouter Bedrock is BYOK-only and EU in-region routing unavailable at the November 10 check

Use Cases

•Difficult coding and repository migration tasks with independent acceptance checks
•Long-horizon agents and multistep research
•Document, spreadsheet and presentation analysis using tools

⚙️Input Parameters

model

string

Claude Messages API: claude-fable-5-2. Provider IDs and schemas are separate.

max_tokens

integer

Required Messages API output budget, up to the documented 128,000-token maximum.

messages

array

Text/image conversation blocks. Preserve returned assistant blocks and append history without editing earlier bound content.

thinking

object

Adaptive is always on. Omit this field or use type adaptive; disabled and manual budget_tokens are unsupported.

output_config.effort

string

low, medium, high, xhigh or max; high is the documented API default.

tools

array

Optional Claude API tool definitions. Forced tool_choice any/tool is unsupported; use auto or none.

💡Usage Examples

Example 1

Input Parameters

{
  "model": "claude-fable-5-2",
  "max_tokens": 8192,
  "messages": [
    {
      "role": "user",
      "content": "Review this proposed database migration. Identify data-loss risks, suggest the smallest safe change, and define checks before execution."
    }
  ],
  "thinking": {
    "type": "adaptive"
  },
  "output_config": {
    "effort": "high"
  }
}

Output Results

Illustrative request only; not executed by GetLLMs. Send it to the Claude Messages API with your own authentication and actual migration context. Inspect returned content and usage blocks; verify any proposed change independently before execution.

Quick Actions

Technical Specifications

Hardware Type: Hosted Claude API and documented partner platforms; no self-hosted checkpoint verified here.
Commercial Use: Supported
Pricing: Standard Claude API USD/M tokens: $8 input, $40 output; $0.20 cache read, $10 5m/$16 1h cache write. Fast mode $16/$80 on supported platforms.