Claude Fable 5

Labellerr's June 10, 2026 review directly compares Claude Fable 5 with Claude Mythos 5, noting both share one trained base and that Fable 5 is generally available while Mythos 5 is restricted to a trusted-access group including cyberdefense partners and biology researchers. The explainer breaks Fable 5's deployment con The article states that internal evaluations show Fable 5's cybersecurity classifiers block meaningful progress on attack tasks, that an external bug-bounty program produced zero universal jailbreaks across more than 1,000 hours of testing, and that in over 95% of sessions no fallback to Opus 4.8 occurs. It places Myth

Coursiv's guide states that Anthropic released Claude Fable 5 on June 9, 2026 as its first generally available Mythos-class model, a new tier above Opus aimed at demanding reasoning, long-horizon agentic work, coding, knowledge work, vision, and large-context analysis. According to this consolidated reference, access w The same guide describes Fable 5 as a "covered model" with a 30-day data retention requirement (not zero data retention) and notes that adaptive thinking is always on while raw chain-of-thought is not returned but summarized thinking can be requested, listing autonomous coding, long-document analysis, complex research,

TrueFoundry's June 10, 2026 explainer documents Claude Fable 5 as the first Mythos-class model Anthropic released for general use, sitting above the Opus family and targeting ambitious long-running tasks such as large-scale code migrations, multi-day agentic sessions, deep research, and dense knowledge work. It repeats The piece positions Fable 5 in Anthropic's lineup as Haiku for speed, Sonnet for balance, Opus for hard problems, and Mythos-class for "the genuinely brutal ones," emphasizing that on short one-shot queries the difference versus prior Claude tiers may be modest but that on multi-hour and multi-day tasks Fable 5's lead

On June 9, 2026, Anthropic released Claude Fable 5, the first model in its new "Mythos-class" tier above Opus, cleared for general availability. Per the Vellum explainer that traces its figures back to Anthropic's launch materials, Fable 5 posts a leading 80.3% on the SWE-Bench Pro agentic-coding benchmark, ahead of Op The same breakdown highlights Fable 5's knowledge-work and vision performance, noting a leading score on Hebbia's senior-level Finance Benchmark with gains in document reasoning, chart analysis, and dense multi-source synthesis, plus token efficiency improvements relative to earlier Claude generations. Anthropic frames

Anthropic announced Claude Fable 5 on June 9, 2026, describing it as a Mythos-class model — a new tier above Opus — and the most capable model it has ever made generally available. The page states Fable 5 is state-of-the-art on nearly all tested benchmarks, with particular strengths in software engineering, knowledge w The same Anthropic page documents a subsequent access disruption: on June 12, 2026, Anthropic suspended access to Claude Fable 5 and Claude Mythos 5, and on July 1, 2026, both models were redeployed and made available again. The announcement frames the capability jump in the context of applied use cases, noting that Fa

OpenRouter's aggregator page lists Claude Fable 5 (model ID claude-fable-5) with the metadata also reflected on Vertex: a 1M-token context window, $10 per million input tokens and $50 per million output tokens, cache reads at $1 per million, text/image/file inputs with text output, reasoning support, and a June 9, 2026 OpenRouter positions the model as a Mythos-class Anthropic offering for autonomous knowledge work and coding, suited to long-running, complex, and asynchronous tasks that previously required frequent human check-ins. It supports BYOK routing across multiple cloud endpoints including Google Vertex, with the data-retenti