Models
The new mid-tier Claude nearly matches the priciest one at half the cost, Anthropic says
Claude Sonnet 5.5 keeps its predecessor's price and, according to Anthropic, comes very close to Opus 5.5, which costs twice as much. Almost all the figures behind that claim come from the company itself.
The essentials
2 confirmed facts · 2 according to the source · 2 open questions
- Anthropic released Claude Sonnet 5.5 in late September 2026, the second model in its 5.5 family after Claude Opus 5.5.134
- It costs the same as Sonnet 5: $2 per million input tokens and $10 per million output tokens, half the price of Opus 5.5.124
- According to the sourceAnthropic says it responds more than 30% faster than Sonnet 5 and that each task costs up to 30% less.123
- According to the sourceAccording to programmer Simon Willison, Sonnet 5.5 is now the model used by the free version of Claude.2
Why it matters
The gap between the mid-tier Claude and the expensive one is narrowing, but the expensive one still costs twice as much. If Anthropic's figures hold up in real-world use, the mid-tier model would be enough for everyday work for many companies and programmers. That it is also, according to Simon Willison, the model in the free version suggests Anthropic is competing for people who don't pay as well2.
If you use Claude without paying, this model is already answering you, according to Simon Willison2. If you pay per use, the price is the same and the savings Anthropic promises depend on each task.
The details
Almost all the performance figures come from Anthropic, MarkTechPost warns. On OSWorld 2.1, which measures how it operates a computer, Sonnet 5.5 scores 80.1% and Opus 5.5 scores 81.8%. On GDPval-AA v2.1, a test run by Artificial Analysis, it scores 1,844 points against 1,846 for Opus 5.51. Anthropic acknowledges that Opus 5.5 is still clearly better at complex, open-ended work1.
The savings do not come from a price cut. According to Anthropic, the model uses fewer tokens, the chunks of text that usage is billed by, and that is why each task costs less13. It is available on Anthropic's platform and on AWS, Google Cloud and Microsoft Azure1. MarkTechPost and The Rundown AI note that it is the first Sonnet with cybersecurity safeguards14.
What we don't know
- Whether the speed and cost improvements Anthropic announces hold up in real-world use: most of the tests are the company's own.
- When Haiku 5.5 will arrive and how much it will cost; Anthropic places it “in the coming weeks”, according to Simon Willison.
Background
On 2 October we reported that OpenAI promises something similar with GPT-6.1 Sol: nearly the performance of its most expensive model for a fifth of the price, also backed by its own tests.
Sources
- 1Anthropic Releases Claude Sonnet 5.5: 70.6% on Terminal-Bench 4.0 at the Same $2/$10 Price
- 2Claude Sonnet 5.5
- 3Claude Sonnet 5.5 🧠, Anthropic IPO leaks 📝, AMD buys World Labs 💰
- 4Anthropic's mid-tier Claude climbs the rankings
- 5Sonnet 5.5 is worth a try
Was this story useful? Yes Not really
This story was written in Spanish by an artificial intelligence system from the sources listed above, and translated automatically. Before publication, a program checks that every figure and every name appears in the sources, and that the translation keeps the same facts, figures and sources. It has not been reviewed by a person. Spanish original · How we work (in Spanish)