Back in April, Anthropic announced Claude Mythos Preview, a frontier model with state-of-the-art coding capabilities. Due to the cybersecurity implications that would occur due to the availability of such a powerful model, Anthropic made it available to only a select set of companies around the world. The company"s plan was to prepare appropriate guardrails before releasing such a powerful model to everyone.
Now, after nearly two months, Anthropic announced Claude Fable 5, its most capable AI model yet for general users. The company also announced Claude Mythos 5, the same underlying model as Fable 5, but with safeguards lifted, making it more suitable for selected cybersecurity and biology use cases.
Claude Fable 5 sits a tier above its Opus models and it beats most other generally available models across areas including software engineering, knowledge work, vision, scientific research, and long-running autonomous tasks.
To prevent model misuse, when Claude Fable 5 detects certain requests related to cybersecurity, biology, chemistry, or model distillation, the request will be routed to the Claude Opus 4.8 model. Anthropic claims that these safeguards trigger in less than 5% of sessions on average.
However, for large organizations working on critical software, Claude Mythos 5 can be availed through Project Glasswing. Later, Anthropic has plans to expand access through a broader trusted access program.
As you can notice in the benchmarks above, Fable 5 and Mythos 5 are state-of-the-art on most key AI benchmarks and they are well ahead of OpenAI"s frontier model, GPT-5.5. For example, Fable 5 is the new state-of-the-art model for vision tasks. Also, Mythos 5 has the strongest cybersecurity capabilities of any model in the world.
Claude Fable 5 and Claude Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, which is less than half the price of Claude Mythos Preview.
Another big change is that Anthropic is making a change to the way they handle business customer data for both Fable 5 and Mythos 5 models. The company will now require 30-day retention for all traffic on both first- and third-party surfaces. Anthropic promises that it won"t use the data to train Claude models, instead it will use it against complex and novel attacks.
Claude Fable 5 is available today on the Claude API and consumption-based Enterprise plans. It is also included at no extra cost for Pro, Max, Team, and seat-based Enterprise customers from today through June 22. After that, users on those plans will need usage credits to continue using Fable 5, unless Anthropic extends the included access window based on capacity. Developers can access Fable 5 through the Claude API using the claude-fable-5 model name.