Anthropic has officially introduced Fable 5.2 and Mythos 5.2, the latest iterations of its advanced AI models engineered to tackle ambitious projects and complex technical problems. Both models arrive with a notable reduction in pricing, offering a 20 percent cheaper token cost compared to previous versions. According to the company, these releases are finely tuned for open-ended technical work, scientific research, and advanced mathematics, rather than serving as routine daily assistants.
Fable 5.2 is currently available for all users, combining improved inference speeds with enhanced cybersecurity and biology safeguards. Meanwhile, access to the broader capabilities of Mythos 5.2 requires users to apply directly through Anthropic’s Life Sciences and Cyber Verification Programs. To showcase the sheer power of Fable 5.2, an Anthropic engineer tasked the model with designing a prototype operating system from scratch. Over the course of eight days, more than 1,000 autonomous agents wrote, tested, and reviewed code, culminating in a functional operating system complete with a boot sequence and integrated Claude functionality.
Key takeaways
- Fable 5.2 and Mythos 5.2 offer token costs that are 20 percent cheaper than previous versions.
- Standard pricing sits at $8 per million input tokens and $40 per million output tokens.
- An optional fast mode increases token generation speed by up to 2.5 times at doubled pricing.
- Fable 5.2 formalized the proof of Fermat’s Last Theorem within 30 hours during testing.
Pricing models and optional fast mode
Standard usage for Fable 5.2 is priced at $8 per million input tokens and $40 per million output tokens, marking a significant cost reduction from the 5.1 release. For developers and researchers working against tight deadlines, Anthropic is introducing an optional fast mode. This setting boosts token generation speeds by up to 2.5 times, though it comes at an increased cost of $16 per million input tokens and $80 per million output tokens. Across nearly all benchmark tasks, the company reports that Fable 5.2 achieves higher accuracy than Opus 5.5 while simultaneously reducing the token cost per task.
Despite the advanced capabilities of the new releases, Anthropic advises taking a balanced approach to model deployment. The company recommends using Opus 5.5 to initiate daily tasks, bringing in Fable 5.2 as a specialized advisor when tackling more demanding obstacles. Tom Pritchard, a staff engineer at Shopify, noted that Fable 5.2 excels at taking loose concepts and refining them into reliable execution plans with minimal human intervention.
Scientific research and complex problem solving
Beyond operating system development, Anthropic demonstrated the versatility of Fable 5.2 across various scientific domains. During internal testing, the model successfully formalized the proof of Fermat’s Last Theorem in a span of 30 hours. Additionally, it independently designed a home voice-assistant hub, a USB speakerphone, and a handheld virtual pet from scratch, alongside solving two open problems in black hole physics. These milestones highlight the model's aptitude for high-level scientific and mathematical research.
Comments