Anthropic launches Claude Fable 5.2 and Mythos 5.2, with the former open to everyone
The company announces two versions of the same model: Fable 5.2, with safeguards and available to all, and Mythos 5.2, only for verified organisations.
On 9 November Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2, which it describes as its most capable models for ambitious, open-ended technical work. Several members of the groups shared the announcement, and one of them said it was already available in the Claude app.
According to the company, they are two versions of the same model. Fable 5.2 includes safeguards against misuse in cybersecurity and biology and is available to all users, including on the cloud services of Amazon, Google and Microsoft; in the API it is called `claude-fable-5-2`. Mythos 5.2 has fewer safeguards and is reserved for organisations that pass its verification programmes in cybersecurity or life sciences.
Anthropic states that they produce results up to 30% faster than the 5.1 versions, that they outperform Opus 5.5 in almost every capability it has evaluated and that, in initial tests, the safety classifiers block 55% less than with Fable 5.1. These are the company's own figures, with no independent verification. The standard price given is $8 per million input tokens and $40 per million output tokens.
For teachers who program with agents, the practical news is that the company's most powerful model is within their reach. A trial open until the 16th was mentioned in the conversation; the announcement does not give that date, only a launch promotion “this week” with additional Fable usage for eligible paid subscribers, so the deadline could not be confirmed.
Anthropic launches Claude Fable 5.2 and Mythos 5.2, with the former open to everyone
The company announces two versions of the same model: Fable 5.2, with safeguards and available to all, and Mythos 5.2, only for verified organisations.
On 9 November Anthropic announced Claude Fable 5.2 and Claude Mythos 5.2, which it describes as its most capable models for ambitious, open-ended technical work. Several members of the groups shared the announcement, and one of them said it was already available in the Claude app.
According to the company, they are two versions of the same model. Fable 5.2 includes safeguards against misuse in cybersecurity and biology and is available to all users, including on the cloud services of Amazon, Google and Microsoft; in the API it is called `claude-fable-5-2`. Mythos 5.2 has fewer safeguards and is reserved for organisations that pass its verification programmes in cybersecurity or life sciences.
Anthropic states that they produce results up to 30% faster than the 5.1 versions, that they outperform Opus 5.5 in almost every capability it has evaluated and that, in initial tests, the safety classifiers block 55% less than with Fable 5.1. These are the company's own figures, with no independent verification. The standard price given is $8 per million input tokens and $40 per million output tokens.
For teachers who program with agents, the practical news is that the company's most powerful model is within their reach. A trial open until the 16th was mentioned in the conversation; the announcement does not give that date, only a launch promotion “this week” with additional Fable usage for eligible paid subscribers, so the deadline could not be confirmed.
