Claude Sonnet 5.5 rivals Opus with 'thinking budget' settings
Claude Sonnet 5.5, released in July, now matches Opus 5.5 in performance on many tasks while costing about half as much. The model’s new 'effort' or 'thinking budget' settings—ranging from low to max—let users adjust output quality and cost dynamically, with high settings sometimes outperforming Opus at medium effort. The author, who previously assigned cheaper tasks to Sonnet and complex work…
What you can do with it
AI at Work →Adjust task quality and cost with 'thinking budget' settings in Claude
What you get You can draft content with tighter control and cut Opus usage by 30%.
Lets users dial up or down Claude’s reasoning depth.
- Who for
- marketers
- Cost
- included in a tool you already have
- Effort
- minutes
- Already included in
- Claude’s subscription plans
Use it for
- Draft content with tighter control
- Cut Opus usage by 30%
- Test prompts side-by-side
How to set it up · From the article
- Set Claude’s effort level to *high* or *medium* for Sonnet/Opus
- Run the same prompt on both models
- Compare token usage and output quality
Watch out Check the details on the maker's page first
Key points from the news
- Claude Sonnet 5.5 matches Opus 5.5 performance in tests while costing ~half as much per million tokens
- 'Thinking budget' settings (low to max) let users trade cost and quality, with *high* often outperforming Opus at *medium*
- Author shifts workflow to use Sonnet with elevated effort first, reducing reliance on pricier Opus or Fable tiers
The author notes that Sonnet 5.5’s ability to handle tasks like playing Pokémon Red via screenshots—something previous Sonnet versions struggled with—demonstrates a leap in capability. The 'thinking budget' setting can drastically alter costs: max effort may quadruple token usage compared to medium, impacting billing. The author’s workflow now prioritizes Sonnet with elevated effort settings to stretch usage limits, especially on the $20/month plan, challenging the notion that serious Claude use requires the $200/month top-tier plan.
Model pages: Claude Sonnet 5.5 → · Claude Opus 5.5 →
How to use Claude Sonnet 5.5 alongside Opus: Adjusting costs and results with 'thinking budget' settings
note.com · 29 September 2026
Loading the full article…
This text was published by note.com and written by sasuu｜非エンジニアのAI活用術. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Marketing & Small Business
All →- ChatGPT research tips cut back-and-forth and verification time · 4 src
- ChatGPT Images 2.0: manuals work better without human hands · 1 src
- Marissa Mayer’s Dazzle turns camera rolls into personalized AI suggestions · 1 src
- Braze launches AI tools for campaign decisioning and external LLM access · 1 src
- ChatGPT adds flashcards for study with basic prompts · 1 src
Comments
via GitHub Discussions