All posts
AI / News

Claude Fable 5.2 Arrives at $8 per Million Input Tokens, With Medium Effort as the New Default

Anthropic: Claude Fable 5.2 and Mythos 5.2 launch

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on 9 November 2026. They are the same underlying model. Fable carries stronger cybersecurity and biology safeguards, which is what lets Anthropic offer it to everyone, while Mythos goes to vetted organisations through its cyber and life sciences verification programmes.

Fable 5.2 is available on the Claude apps, the API as claude-fable-5-2, and on AWS, Google Cloud and Microsoft Azure. Standard pricing is $8 per million input tokens and $40 per million output tokens. An optional fast mode costs $16 and $80 and, according to Anthropic, speeds up token generation by up to 2.5 times.

Even without fast mode, Anthropic says output is up to 30 percent faster than Fable 5.1. The other change people will notice is the default: Medium is now the effort setting in the Claude apps and Claude Code, and Anthropic says 5.2 completes hard tasks with fewer tokens than 5.1 did at that level.

The benchmark table is long, so we will pick the numbers that matter to a studio like ours. On FrontierCode 2.0 Fable 5.2 scores 49.1 percent at its highest effort setting. At High effort it lands level with Opus 5.5 (45.3 against 45.5 percent) at roughly half the cost per task. On Chartography, a chart reading test, it scores 73.4 percent against 64.4 percent for Opus 5.5.

Anthropic also reports about 55 percent fewer safeguard blocks than Fable 5.1 in early testing. Customers without zero data retention get 30 day retention for safety monitoring, which is worth checking before you push client material through it.

Interestingly, Anthropic's own page still recommends Opus 5.5 as the daily driver for most coding and knowledge work, and positions Fable for the most ambitious, open-ended technical jobs. That matches how we would use it. Day to day copy, admin and small web fixes stay on the cheaper model.

Where Fable earns its price for us is long, messy builds: refactoring a 3D configurator, reworking a CMS content model, or untangling a codebase we have inherited. The chart reading gain is also useful for the reporting side of client work, where a lot of the input arrives as screenshots of dashboards rather than clean data.

If you have Fable wired into a pipeline already, check two things this week. First, whether the new Medium default changes the depth of answers you were relying on. Second, whether your integrations are affected by preserved thinking, which Anthropic says now applies to all accounts.