4 September 2026

Anthropic releases Claude Fable 5.1 and Mythos 5.1 models

First reported

Deep Learning Weekly ran this on .

  • Anthropic, the company behind the Claude chatbot, released two updated Claude models with performance improvements.
  • Cache reads, a way to reuse previously processed text and save costs, dropped to $0.25 per million tokens with a 75% reduction in pricing.
  • Terminal-Bench-Science scores, a measure of how well models handle scientific reasoning tasks, more than doubled to 52.6%.

How it was covered

Deep Learning WeeklyEditorial team

Anthropic released two new Claude models with significant improvements: cache reads reduced by 75% to $0.25 per million tokens and Terminal-Bench-Science scores more than doubled to 52.6%.