Anthropic released Claude Fable 5.1 on September 1, a model achieving 90% coverage on the ARC-AGI-2 benchmark compared to 89.2% for Fable 5, at a verified cost of $5.45 per task. The model doubled its performance on scientific tasks with a score of 52.6% on Terminal-Bench-Science 0.1 versus 24.7% for its predecessor. Anthropic cut its cache-read pricing by 75% to $0.25 per million tokens, generating savings of 25% to 45% depending on workflow complexity. Base pricing remains unchanged at $10 per million tokens for standard tasks and $50 for complex tasks. The model features a 1 million token context window and 128K maximum output.
Source: Read the original article
