Anthropic's Claude Mythos exits Preview on Google Cloud, expanding enterprise access

Anthropic’s Claude Mythos has quietly shed its 'Preview' label on Google Cloud’s Vertex AI, a change spotted by AI researchers monitoring the console’s model listing. The specific entry for 'base-model-claude-mythos' now appears without the preview tag as of May 17, 2026. This pattern — removing the label before a broader rollout — was previously observed with Claude Opus 4.7, which subsequently became widely available to enterprise customers on Vertex AI. Anthropic has yet to make an official statement, but the infrastructure change strongly indicates that Mythos is being primed for expanded access across Google Cloud’s enterprise ecosystem.

Claude Mythos is Anthropic’s most advanced large language model to date, designed for complex reasoning, long-context tasks, and high-stakes enterprise use cases. Its transition from preview to general availability would allow organizations to deploy it in production without the typical preview limitations (e.g., rate caps, no SLAs). For Google Cloud customers already using Opus 4.7 or Claude 3.5 Haiku, adding Mythos would provide a tier for ultra-complex workflows like multi-step agent reasoning, financial modeling, or advanced code generation. The move also strengthens Anthropic’s position against OpenAI’s GPT-5 and Google’s Gemini Ultra, especially in regulated industries that require verifiable, safe AI outputs. Enterprise teams should monitor for an official launch announcement and pricing adjustments in the coming weeks.