Claude Fable 5 Released: Benchmarks, Pricing, Availability, and Real-World Examples
Anthropic has released Claude Fable 5, the first general-use “Mythos class” model, and the video reviews the announcement blog post, reactions, benchmarks, pricing, availability, and examples. Fable 5 is described as state-of-the-art across nearly all tested benchmarks, with notable strengths in agentic coding, knowledge work, vision, and scientific domains like biology and health, and larger gains on longer, more complex tasks. Pricing is $10/M input tokens and $50/M output tokens, with web access via Pro/Max tiers but limited availability until June 22 and possible metered costs even for some subscribers. The video highlights Frontier Code “no-slop code” results and tradeoffs between effort level, cost, and performance, plus anecdotes like completing Pokémon FireRed from screenshots alone and demos including an HTML solar system simulation and natural-language CAD (VibeCAD). It also covers access via Project Glasswing for Mythos 5, Claude Code/Managed Agents, safety tuning and refusals, a 319-page model card, and usage tips emphasizing simpler prompting and managing iterative “loops.”
Model Card: https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c342ee809620.pdf
00:00 Fable 5 Arrives
00:22 Benchmark Gains and Strengths
01:35 Pricing and Subscription Window
02:26 Frontier Code No Slop Results
03:29 Pokemon and Visual Demos
04:41 Access and Safety Notes
05:49 How to Use It Better
06:55 Loops and Final Benchmark
07:34 Wrap Up and Call to Action
source
