Anthropic launches public Mythos model with safeguards, limits high-risk access

Anthropic launches public Mythos model with safeguards, limits high-risk access

The release intensifies competition in advanced AI models as Anthropic targets a strong Arena Leaderboard showing while keeping restrictions on sensitive uses such as cybersecurity.

Summary

Anthropic launched a public version of its Mythos model, widening access while maintaining restrictions on high-risk uses such as cybersecurity. The release adds competitive pressure in the AI model race and could influence strategic moves across the sector, with Anthropic also signaling it is aiming for a high debut on the Arena Leaderboard. Earlier launch expectations had risen on Polymarket, while the company has described Claude Fable 5 as its strongest broadly deployable model and priced Mythos preview and Claude Mythos 5 at $10 per million input tokens and $50 per million output tokens.

Terms & Concepts
  • Arena Leaderboard: A ranking system used to compare AI models based on performance, often shaping industry perception and competitive positioning.
  • Polymarket: A prediction market where users trade on event outcomes, producing market-implied probabilities.
  • input tokens: Units of text or data sent into an AI model for processing and billing.