Anthropic introduced Claude Fable 5.1 and Claude Mythos 5.1 on September 1. They use the same underlying model with different safeguards. Fable is generally available; Mythos is limited to trusted access programs.
Fable 5.1 supports a one-million-token context window and improves coding and sustained knowledge work. Its cache-read price is 75% lower than Fable 5. Anthropic estimates about 25% savings for typical token-billed workloads, with larger savings possible for agent workflows that reuse substantial context.
The company also reports 60% fewer cybersecurity safeguard false positives. That means fewer benign requests are blocked; it does not mean the model makes 60% fewer coding mistakes. New API accounts have restrictions on editing earlier assistant messages in multi-turn conversations as part of anti-distillation protections.
Source: Anthropic's Fable and Mythos 5.1 announcement.
Adapted from the original Chinese article, published on September 2, 2026.
Comments NOTHING