Anthropic puts a new top model on the field with Claude Fable 5.2

Anthropic has a new starter. On 9 November it released Claude Fable 5.2 to its API, to its apps for Pro, Max, Ultra, Team and Enterprise users, and to AWS, Google Cloud and Microsoft Azure. It also released Claude Mythos 5.2. Anthropic says the two are the same underlying model. Fable carries strong cybersecurity and biology safeguards, which Anthropic says lets it go to all users. Mythos is open only to vetted cybersecurity and life sciences groups.

Think of Mythos as the same player kept on a closed practice squad. Fable is the version cleared to play in public games. Anthropic says many of its cybersecurity and biology queries are automatically routed to less capable models.

What Anthropic claims: its own launch table shows its new 5.2 models ahead of Claude Opus 5.5 on almost every test it lists. Those are the company's numbers. What has been measured outside: Artificial Analysis, an independent tester, scores Fable 5.2 at 65 at its highest effort setting on its Intelligence Index, 1st of 74 models it lists. The Arena human vote board does not list it yet.

For the AI models division of the Power Rankings, Anthropic's entry today is Claude Opus 5.5 in 1st place. Our model places come mainly from independent tests, so Friday's run will decide how Fable 5.2 is counted.

What would change this read: Arena votes, more independent tests, or a wider Mythos rollout.

Rankings and game results describe what already happened. They are not a view on where any share price goes next. This is general information, not investment advice.