Anthropic Just Made Mythos Public, and Called It Fable 5
Anthropic took the model that spooked the U.S. government and shipped a version of it to anyone with a Pro subscription. Fable 5 is a Mythos-class model, the same underlying weights as the new Mythos 5, but with safeguards bolted on so it can go out to the public. In high-risk zones like cybersecurity, biology, and chemistry, it just refuses and quietly falls back to Opus 4.8. Everywhere else, it's the most capable thing Anthropic has ever let regular users touch.
The number that matters: Stripe ran it on a codebase-wide migration that would have taken a team over two months by hand. Fable 5 did it in a day. That's the whole pitch for agentic coding in one sentence, not "it writes better functions" but "it compresses months into days." It's state-of-the-art on nearly every benchmark they tested: software engineering, knowledge work, vision, science.
Pricing is $10 per million input tokens and $50 per million output. Not cheap, but through June 22 it's free inside Pro, Max, Team, and seat-based Enterprise plans. Anthropic clearly wants everyone to feel the jump before the meter starts. Mythos 5, the unrestricted sibling, goes only to pre-approved orgs and has the strongest cybersecurity capabilities of any model on earth, which is exactly why the public doesn't get it.
The interesting tension: Anthropic spent the last two weeks warning that AI is getting too dangerous, then released its most powerful public model days later. The reconciliation is the safeguard layer. They're betting that block the bio and cyber stuff and fall back is enough to ship frontier capability safely. Whether that holds is the story of the next year. Link: https://www.anthropic.com/news/claude-fable-5-mythos-5
← Back to all articles
The number that matters: Stripe ran it on a codebase-wide migration that would have taken a team over two months by hand. Fable 5 did it in a day. That's the whole pitch for agentic coding in one sentence, not "it writes better functions" but "it compresses months into days." It's state-of-the-art on nearly every benchmark they tested: software engineering, knowledge work, vision, science.
Pricing is $10 per million input tokens and $50 per million output. Not cheap, but through June 22 it's free inside Pro, Max, Team, and seat-based Enterprise plans. Anthropic clearly wants everyone to feel the jump before the meter starts. Mythos 5, the unrestricted sibling, goes only to pre-approved orgs and has the strongest cybersecurity capabilities of any model on earth, which is exactly why the public doesn't get it.
The interesting tension: Anthropic spent the last two weeks warning that AI is getting too dangerous, then released its most powerful public model days later. The reconciliation is the safeguard layer. They're betting that block the bio and cyber stuff and fall back is enough to ship frontier capability safely. Whether that holds is the story of the next year. Link: https://www.anthropic.com/news/claude-fable-5-mythos-5
Comments