Anthropic: Mythos 5 and Fable 5 Revolutionize AI at a High Price

Anthropic, the San Francisco-based startup, recently launched its most advanced artificial intelligence models: Claude Fable 5 and Mythos 5. Announced on June 9, these models represent a significant advancement in terms of intelligence and safety, but they also come with a notable increase in pricing.

Claude Fable 5, derived from the Mythos range, stands out for its exceptional performance in areas such as software engineering, knowledge work, and visual analysis. It outperforms its competitors in complex tasks requiring a high degree of autonomy. Benchmark results are impressive, with 80.3% on SWE-Bench Pro and 88.0% on Terminal-Bench 2.1, surpassing Opus 4.8 and other models like GPT 5.5 and Gemini 3.1 Pro. In terms of agentic code, Fable 5 achieves 29.3% on FrontierCode, compared to 13.4% for Opus.

Anthropic has integrated specific safeguards into Fable 5 to prevent risks of misuse, particularly in the fields of biology and cybersecurity. These safeguards redirect queries to Claude Opus 4.8. This measure comes amid growing concerns about the biological risks associated with advanced AIs. A public letter co-signed by industry leaders has called for strict controls on the sale of synthetic DNA and RNA.

Concrete discoveries made with Mythos 5 include a tenfold acceleration in drug design, with nine promising candidates out of 14 targets, and a novel hypothesis about an E. coli protein confirmed by a third-party team. A genomics model has also been trained in near-autonomy.

On the pricing front, Claude Fable 5 and Mythos 5 are priced high: $10 per million tokens for input and $50 for output. In comparison, Claude Opus 4.8 is offered at $5 for input and $25 for output. This increase reflects the enhanced sophistication of the models.

Fable 5 is available via API and several platforms, with no additional cost for Pro, Max, Team, and Enterprise subscriptions until June 22. After this date, a usage credit system will be implemented. Anthropic aims to solidify its position in the premium AI market, targeting companies and laboratories that require raw intelligence and maximum autonomy. Data retention is limited to thirty days, with no training, and access logging, an important detail for user security.

With Mythos 5 and Fable 5, Anthropic asserts its position as a premium AI publisher, reserving these models for specific uses that require maximum intelligence and autonomy, such as code agents, complex analyses in finance, legal or consulting fields, and AI R&D.