Source: the-decoder.com/anthropic-says-outdated-prompting-habits-are-slowing-down-claude-opus-5-5
Anthropic says users of Claude Opus 5.5 should drop prompting habits carried over from older model generations. Its top advice is to remove phrases like "think hard" or "think carefully," because the model always reasons on its own and scales its reasoning depth to the task and the selected effort level. In Anthropic's tests, responses without these phrases started faster with no measurable drop in quality.
Anthropic also recommends starting at the Medium effort level, which is the default for Opus 5.5. The company says that in internal tests, Opus 5.5 on Medium matched its predecessor Opus 5 on High, and it advises switching to High only when Medium is not enough. The setting can be changed in the model menu next to the send button.
For multi-step work, Anthropic suggests describing the desired result in one message, covering the goal, output format and when the model should ask questions. Diagrams and tables should be attached as images or files instead of typed out, and design requests should name unwanted styles specifically rather than banning generic design. The company also argues that teams who reserved Opus for complex projects and used Sonnet for the rest should reconsider, since Opus 5.5 can finish tasks in fewer steps.
Why it matters if you build agents
Agent builders who keep reasoning-cue phrases or step-by-step prompt chains may be adding latency without gaining output quality on Opus 5.5.
Choosing a lower effort level as the starting point could reduce cost per task, and per Anthropic, a single well-specified request can replace several back-and-forth turns.
Get stories like this every weekday
One short email with the day's agent tooling news.