Claude Opus 5.5 Now Default Across Paid Tiers

Anthropic has quietly flipped the switch: Claude Opus 5.5 is now the default model across all paid tiers — Pro, Max, Ultra, and Team, and it’s behaving nothing like the Opus users trained themselves to prompt over the last 18 months. The company says outdated habits are throttling it: phrases like “think carefully”, effort-level overkill, step-by-step chat choreography, and manual data transcription aren’t just unnecessary, they’re actively slowing responses down, with no upside in quality.

That’s the core of Anthropic’s new guidance, published Thursday afternoon and aimed squarely at users who’ve carried forward prompting muscle memory from Opus 4 or earlier. It’s not a feature drop or a price change. It’s a usability reset, one that assumes you’re already paying for Opus, and now need to stop holding it back.

And it arrives as teams re-evaluate cost-per-task trade-offs: Anthropic claims Opus 5.5 on Medium effort matches Opus 5 on High, and that cranking Sonnet to Extra High or Max often costs as much as running Opus 5.5 outright. So if your team still treats Opus as a rare, high-stakes tool while leaning on Sonnet for everything else, Anthropic wants you to test the reverse.