The watermark is coming to Claude Mythos 5, Opus 4.7, Opus 4.6, and Sonnet 4.6. Anthropic says changing any one of temperature, top_p, or top_k is enough to bypass it on Opus 4.6 and Sonnet 4.6; it advises keeping those settings at their defaults for compliance.

Anthropic says the watermark does not change meaning, quality, or readability, though that claim is disputed. The company says the rollout prepares older models for the EU AI Act’s December deadline.

Why it matters

API users relying on watermarks for compliance need to account for the fact that changing default sampling settings may remove the mark on two models.

GUIDES

Something to try

Hypothetically, if your compliance process relies on watermarks, audit which models and sampling settings your systems use.