Anthropic unveils Claude Fable 5 and Mythos 5 with expanded capabilities
Anthropic has launched two new models under its Mythos-class tier—Claude Fable 5 and Claude Mythos 5—both built on the same underlying architecture but designed for different use cases. Released on June 9, 2026, Fable 5 is positioned as the safer, widely available option, while Mythos 5 removes some safeguards for advanced applications, remaining in limited release.
A tiered approach to capability and control
The Mythos-class sits above Anthropic’s Opus tier in terms of performance, with both new models supporting a default context window of 1 million tokens and allowing up to 128,000 output tokens per request. Pricing reflects their enhanced capabilities: $10 per million input tokens and $50 per million output tokens—less than half the cost of the earlier Claude Mythos Preview. The naming reflects this dual-purpose design: “Fable” draws from fabula, suggesting narrative clarity, while “Mythos” evokes broader, unfiltered exploration.
Performance gains across domains
Anthropic reports that Fable 5 achieves state-of-the-art results across software engineering, knowledge work, vision, and scientific research. In software engineering, early testers like Stripe reported that Fable 5 completed a full codebase migration in a 50-million-line Ruby project in a single day—work that would take a human team over two months. On Cognition’s FrontierCode benchmark, it led all frontier models even at medium effort levels.
In vision tasks, Fable 5 demonstrated the ability to extract precise data from complex scientific figures and reconstruct web apps from screenshots with minimal guidance. It also excelled in long-context reasoning, maintaining focus across millions of tokens and improving outputs by integrating its own notes—a feature that delivered three times the performance boost in a persistent-memory task compared to Opus 4.8.
Mythos 5, while sharing the same foundation, is tailored for research and exploratory use. Anthropic notes internal protein design experts accelerated parts of drug discovery by up to ten times using Mythos 5. In blinded evaluations, scientists preferred its molecular biology hypotheses 80% of the time, and it autonomously conducted genomics research, training a custom model on single-cell data spanning 138 animal species over a week.
Source: MarkTechPost. AI-assisted editorial synthesis — TechnoExpress.