Anthropic launches Claude Fable 5, its most powerful public model
Anthropic released Claude Fable 5 on June 9, 2026, describing it as the most powerful model it has ever made generally available, with vendor-reported benchmark scores including 80.3% on SWE-Bench Pro and pricing at $10/$50 per million input/output tokens.
Score breakdown
Fable 5 represents Anthropic's first public release of a Mythos-class model, with notably higher vendor-reported coding benchmark scores than prior models, and introduces an automatic safety fallback that routes the riskiest request categories to a different model entirely.
- 01Anthropic released Claude Fable 5 on June 9, 2026, calling it the most powerful model it has ever made generally available.
- 02Fable 5 is described as a 'Mythos-class model made safe for general use' — a guardrailed version of the less-restricted Mythos 5.
- 03Vendor-reported SWE-Bench Pro score is 80.3%, compared to 69.2% for Opus 4.8, 58.6% for GPT-5.5, and 54.2% for Gemini 3.1 Pro.
Anthropic released Claude Fable 5 on June 9, 2026, positioning it as the most powerful model it has ever made generally available. The post explains that two models actually exist: Mythos 5, a full frontier model with fewer restrictions, and Fable 5, the publicly available version built on the same core intelligence but with stronger safety guardrails. Anthropic describes Fable 5 as a "Mythos-class model made safe for general use."
The post cautions that these are self-reported numbers and independent evaluations will be needed to validate the claims.
Anthropic's vendor-reported benchmarks show Fable 5 scoring 80.3% on SWE-Bench Pro (versus 69.2% for Opus 4.8, 58.6% for GPT-5.5, and 54.2% for Gemini 3.1 Pro), breaking 90% on "core long-task analytics" (roughly 10 points above Opus 4.8), 46.1% on BioMysteryBench, and 78.0% on ExploitBench. The post cautions that these are self-reported numbers and independent evaluations will be needed to validate the claims. Anthropic also states that Fable 5's performance advantage grows with task length and complexity, with strengths in software engineering, knowledge work, vision, and scientific research.
Pricing is set at $10 per million input tokens and $50 per million output tokens. Fable 5 is available free to Claude Pro, Max, Team, and Enterprise users during an introductory window from June 9 to June 22, 2026. Developers can access it via the Claude API as `claude-fable-5` and on Amazon Bedrock; legal-AI company Harvey announced availability on launch day. A notable safety mechanism routes requests touching cybersecurity, biology and chemistry, or model distillation to Claude Opus 4.8 instead — a fallback Anthropic says is triggered in under 5% of sessions on average.
Key facts
- 01Anthropic released Claude Fable 5 on June 9, 2026, calling it the most powerful model it has ever made generally available.
- 02Fable 5 is described as a 'Mythos-class model made safe for general use' — a guardrailed version of the less-restricted Mythos 5.
- 03Vendor-reported SWE-Bench Pro score is 80.3%, compared to 69.2% for Opus 4.8, 58.6% for GPT-5.5, and 54.2% for Gemini 3.1 Pro.
- 04Pricing is $10 per million input tokens and $50 per million output tokens.
- 05Fable 5 is free for Claude Pro, Max, Team, and Enterprise users from June 9 to June 22, 2026.
- 06The model is available on the Claude API as `claude-fable-5` and on Amazon Bedrock.
- 07A safety mechanism automatically routes requests in cybersecurity, biology/chemistry, or model distillation to Claude Opus 4.8, triggering in under 5% of sessions on average.
Topics
Summary and scoring are generated automatically from the original article. We always link back to the publisher and never republish images or paywalled content. Last processed Jun 10, 2026 · 15:34 UTC. How this works →