AI news · Anthropic · Claude
Anthropic makes Opus 5.5 the default and says drop “think hard”
Claude’s paid plans now open on Opus 5.5, and Anthropic’s new advice tells users to start at Medium effort and cut prompting habits built for older models.
Claude has a new default model. Claude Opus 5.5 is now the default on the Pro, Max, Ultra and Team plans, Anthropic says in a guide The Decoder covered on Saturday 14 November 2026, and it now recommends Opus over the cheaper Sonnet for everyday research, analysis and writing. A more capable model can finish a task by doing less work, Anthropic argues, which can offset much of its higher price per token, the unit of text a model reads and writes. Opus 5.5 is listed at $4 per million input tokens and $20 per million output tokens, against $2 and $10 for Sonnet 5.5.
Anthropic also wants users to drop old habits, The Decoder reports. Anthropic says the model already scales how deeply it reasons to the task and the chosen effort level, so phrases such as “think hard” add nothing; in its trials, replies began sooner without them, with no measurable loss of quality. Its other tips: start at Medium effort, the default setting for how hard the model works, which matched the older Opus 5 on High in internal checks; describe the end result in one message; attach charts as images. Pushing Sonnet to Extra high or Max effort costs about as much as Opus 5.5, Anthropic adds.
On a subscription plan, the extra cost shows up in usage: Anthropic notes that higher-priced models use up a plan’s limits faster. Our sheet lists Claude Pro at $20 a month, or $17 billed yearly, Max at $100 and Ultra at $200 (checked 15 November 2026, our prices).
Opus 5.5 costs twice Sonnet 5.5 per token and, by Anthropic’s own note, uses a plan’s limits faster, so on a $20 Pro plan the new default pays off only if Anthropic’s fewer-steps claim holds.
Sources
- Anthropic: Which Claude model is best to use for knowledge work. Checked
- The Decoder: Anthropic says outdated prompting habits are slowing down Claude Opus 5.5. Checked
More AI news from
- Amazon plans a World Product Graph, Reuters says
- Italy fines Meta €805 million over Facebook gallery suggestions
- Hyundai Home Shopping airs a sales show made entirely by AI
- Constant Contact’s Catalyst agent runs small-business marketing
- AppsFlyer’s new agent can change Apple Ads bids after approval
- Three AI agents planned Tiger Brokers’ billboards in Singapore
- Edelman finds 84% of B2B buyers summarise vendor content with AI
- Channel Factory adds xpln.ai attention data to its ad buying
- Runway opens its Praxis-1 robot model under Apache 2.0