Anthropic's next-generation AI model, codenamed Capybara and internally referred to as Claude Mythos, has been revealed through an accidental data leak. This new tier is reportedly positioned above Opus and demonstrates significantly improved performance in coding, reasoning, and cybersecurity compared to Claude Opus 4.6, though specific metrics are not yet public. The leak also indicated high serving costs for Claude Mythos, with Anthropic planning further efficiency optimizations before any potential general release. The mention of advanced cybersecurity capabilities in the leaked documents caused a brief but notable dip in cybersecurity stock prices. AI
IMPACT This leak reveals a new, potentially more capable AI model from Anthropic, indicating advancements in coding and reasoning that could influence future AI development and adoption.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Mastodon — mastodon.social →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
