Claude Fable 5: Anthropic Releases First Mythos-Class Model to the Public
Anthropic made a historic leap this week by releasing Claude Fable 5 - the first model from its Mythos line, previously restricted to the closed Project Glasswing program - for general public use.
What is Claude Fable 5?
Fable 5 is essentially the same model as Claude Mythos 5, but with additional safety classifiers that redirect cybersecurity and biology queries to Claude Opus 4.8. Anthropic describes it as "state of the art on nearly every benchmark we tested."
"The longer and more complex the task, the larger Fable 5's lead over our other models," Anthropic states in its official announcement.
Impressive Benchmarks
The numbers are striking and place Fable 5 on a level above any current public model:
- SWE-Bench Pro: 80.3% - vs GPT-5.5's 58.6%
- Terminal-Bench 2.1: 88.0% (Mythos 5)
- Tool use: 17.4% improvement over Opus 4.8
- OSWorld-Verified: 83.4%
- GDPval-AA: 1932 (vs Opus 4.8's 1890)
Independent testers report Fable 5 stays on task longer, validates its own work before declaring completion, and maintains context across significantly longer horizons.
Pricing and Availability
The model arrives with a premium price tag:
- Input: $10/million tokens
- Output: $50/million tokens
- Prompt caching: 90% discount on cached tokens
For Claude Pro, Max, Team, and Enterprise subscribers, Fable 5 is available at no extra cost until June 22, 2026. After that, paid usage credits will be required.
What This Means for Developers
For those working with software engineering, large-scale code migrations, and multi-day AI agent sessions, Fable 5 represents a qualitative leap. Stripe, which tested the model early, reported substantial improvements in software engineering workflows.
Anthropic's message is clear: the era of deep reasoning, autonomous models for the general public has begun.