Fable Model Guide

When was Claude Fable 5.2 released?

Anthropic released Claude Fable 5.2 on November 9, 2026.

The announcement and the model’s release record give the same date. The November 11 edit to the announcement corrected a description of a math demonstration. The exact release time was not published in either source. Announcement · Model release record

Launch channels, model IDs, and fast mode

This comparison matches the provider IDs in Anthropic’s model table against the channels named for fast mode in the launch announcement.

  • Claude API and Claude Platform on AWS: claude-fable-5-2. Anthropic listed fast mode on the Claude Platform. It did not list the AWS-hosted Claude Platform as a separate fast-mode entry.
  • Amazon Bedrock Runtime: global.anthropic.claude-fable-5-2. The model has a Bedrock ID, but Bedrock Runtime is absent from the announcement’s fast-mode list.
  • Google Cloud: claude-fable-5-2 on both the Google-managed and Anthropic-managed routes. The launch announcement names both routes for fast mode.
  • Microsoft Foundry: claude-fable-5-2. The fast-mode entry specifies Anthropic hosting.
  • Claude Code: Anthropic names it as a fast-mode channel. The launch announcement also describes a temporary Fable allowance for Pro and standard Team seats of up to half their weekly usage limits. It gives no fixed message count.

Official model ID table · Launch availability and plan terms

Published rates and access limits

  • Claude Platform base rates were $8 for input and $40 for output per million tokens. The announced fast-mode rates were $16 and $80 respectively. Official pricing · Fast-mode announcement
  • The model overview lists a 1 million token context window and a 128,000-token maximum output. It lists November 9, 2027 as the earliest retirement date, with no exact retirement date published there. Model specifications and availability
  • Anthropic requires 30-day data retention unless it expressly approves zero data retention for the customer. Adaptive thinking stays enabled, and forcing a tool with tool_choice set to any or tool returns HTTP 400. Model restrictions