Key Point: Claude Fable 5 costs twice as much as Claude Opus 4.8, but demonstrates significantly superior performance on complex knowledge queries.
Anthropic has released Claude Fable 5 and Claude Mythos 5. According to the vendor, Fable 5 offers the same capabilities as Mythos 5, but employs enhanced safety safeguards against misuse.
Claude Fable 5 and Claude Mythos 5 are now available. After five hours of testing, Fable 5 proves to be a highly capable model: it processes requests reliably and handles complex tasks without apparent limitations. Computation time is noticeably longer, and costs correspondingly higher than previous Claude models.
The technical specifications: 1-million-token context window, 128,000 maximum output tokens, knowledge cutoff in January 2026. Pricing is set at $10 per million input tokens and $50 per million output tokens — double the prices of Claude Opus 4.8. No additional fees apply for longer contexts. Claude Mythos 5 is described as an identical model without the safety classifiers.
The distinguishing feature lies in the guardrails: Fable 5 has significantly stricter safety safeguards that reject requests more frequently. Anthropic has extended the API with new signaling mechanisms to flag rejected requests. At the same time, a fallback option enables automatic switching to another model when security filters are triggered.
Practical testing revealed a significant performance difference on knowledge questions: while Claude Opus 4.8 clearly demonstrated the limits of its knowledge when queried about open-source projects, Fable 5 provided a more comprehensive list with more precise version numbers and release dates. The model also recognized the typo in the input and corrected it implicitly.
Source: simonwillison.net · Published 10 June 2026
Lumi AI News — AI-assisted curation pursuant to Art. 50 EU AI Act. Paraphrase and classification by Lumi News Pipeline v1.6.5.