Anthropic extends its text watermark to four more Claude models, but changing sampling settings avoids it on two
On Claude Opus 4.6 and Sonnet 4.6, the mark applies only when temperature, top_p and top_k stay at their default values.
Anthropic told API customers by email that it will add its text watermark to four more models: Claude Mythos 5, Claude Opus 4.7, Claude Opus 4.6 and Claude Sonnet 4.6, The Decoder reports. Newer models such as Claude Opus 5.5 already carry the mark, and the rollout on Amazon Bedrock may take a few extra days. Anthropic describes the watermark as an "imperceptible pattern in word choice" that does not change meaning, quality or readability. The company says it adds no characters or tokens, does not affect pricing or request and response formats, and barely slows output. It also says the watermark "contains no information about the user, their organization, or their conversations with Claude." There is a limit on the two older models. On Claude Opus 4.6 and Claude Sonnet 4.6, Anthropic says the watermark is applied only when developers leave the three sampling parameters, temperature, top_p and top_k, at their defaults. These settings control how much randomness the model uses when choosing words. The Decoder notes that a small change to any one of them means output from those two models is not watermarked. That limits how reliably text from those models can be traced back to Claude, while developers who change the settings for ordinary reasons may lose the mark without intending to.