Anthropic splits a new model in two: one open to all, one gated

Fable 5.2 is open to all users at lower list prices; the less restricted Mythos 5.2 goes only to verified security and life-science organizations.

On November 9, 2026, shortly before its IPO, Anthropic released two models built on one foundation: Claude Fable 5.2 and Claude Mythos 5.2. Fable is available to all users but has safeguards that block risky cybersecurity and biology tasks. Mythos removes some of those restrictions and is open only to organizations accepted into Anthropic's Verification Programs for Cybersecurity or Life Sciences.

List prices are $8 per million input tokens and $40 per million output tokens, down from $10 and $50 for the predecessors. Anthropic says output is up to 30 percent faster than Fable 5.1 and Mythos 5.1. Select organizations can get an optional fast mode at up to 2.5x speed for twice the cost. On the Artificial Analysis Intelligence Index v5, Fable 5.2 scores 65 at the top tier, ahead of Opus 5.5 at 59 and the unreleased Gemini 4 Carbon at 60. Anthropic also says about 1,000 Fable 5.2 agents formalized the proof of Fermat's Last Theorem in Lean in 30 hours.

For abundance, the notable part is access. A top-scoring model is open to everyone, and the per-token price is 20 percent lower. The tiered release also shows a way to handle dual-use risk: the most capable security and biology work goes to vetted groups, and the general model keeps guardrails.

The caveats matter. Most performance and cybersecurity figures come from Anthropic's own tests. The cost of a task depends on how many tokens it uses. At peak quality, Fable 5.2 averages $26.50 per task, against $10 for Opus 5.5 and $12.99 for Fable 5.1. Anthropic still recommends Opus 5.5 for everyday use. Fable 5.2 also trails rivals in agentic knowledge work and terminal-based coding, and fast mode is limited to certain platforms.

Source: The Decoder