Anthropic on June 9 announced Claude Fable 5 and Claude Mythos 5, describing Fable 5 as a Mythos-class model “made safe for general use” and positioning Mythos 5 as a version with some safeguards lifted for a limited group of users. The company claims Fable 5 is “state-of-the-art” on nearly all of its tested benchmarks, with stronger results in software engineering, knowledge work, vision, and scientific research, though those performance claims come from Anthropic’s own evaluation materials.
The launch centers on two versions of the same underlying model. Anthropic states that Fable 5 will be the general-release model, while Mythos 5 will initially go to cyberdefenders and infrastructure providers through Project Glasswing, in collaboration with the US government. The company also plans a broader trusted access program later on.
Pricing is set at $10 per million input tokens and $50 per million output tokens, which Anthropic says is less than half the price of Claude Mythos Preview. The company also frames the release as a way to bring “advanced AI capabilities” to more users while trying to manage the risks it associates with high-end cybersecurity and biology capabilities.
Anthropic devotes much of the announcement to benchmark-style claims. It says Fable 5 performed especially well on software engineering tasks, including a reported migration in a 50-million-line Ruby codebase at Stripe, and on Cognition’s FrontierCode evaluation. It also points to gains in finance benchmark work from Hebbia and IMC, along with vision tasks such as extracting numbers from scientific figures and rebuilding web app source code from screenshots.
The company also highlights more speculative-looking examples in memory and science. In its account, Fable 5 handled long-running tasks better than prior Claude models, while Mythos 5 reportedly accelerated protein design, produced novel molecular biology hypotheses, and completed genomics research over more than a week of largely autonomous work. Anthropic adds that one of its hypotheses about an E. coli protein was later corroborated in a bioRxiv study.
Safety remains a central theme. Anthropic says Fable 5 uses new classifiers that can route some cybersecurity, biology and chemistry, or distillation-related requests to Claude Opus 4.8 instead of the main model. The company says those safeguards are intentionally conservative and that more than 95% of sessions do not hit a fallback. It also states that Fable 5 will use 30-day retention for business customer traffic on Mythos-class models, with logging and deletion protections attached.
Availability differs by product and plan. Anthropic says Claude Fable 5 is available everywhere today, while Claude Mythos 5 is limited to Glasswing partners for now and later to selected biology researchers. On the Claude API and consumption-based Enterprise plans, Fable 5 is available from today. For subscription plans, Anthropic says it will be included at no extra cost through June 22 on Pro, Max, Team, and seat-based Enterprise plans, before moving to usage credits on June 23 unless the company extends the window.
Source: Anthropic
