Sonnet 5.5 Keeps Its Price And Cuts The Bill

Same Price List, Lower Bill

Anthropic released Claude Sonnet 5.5 on 28 September 2026, the second model in its Claude 5.5 family after Opus 5.5. It is priced exactly like Sonnet 5 at $2 per million input tokens, $10 per million output tokens and $0.20 per million tokens for cache reads. The saving comes from the model needing far fewer tokens to finish the same work.

These are Anthropic's own figures, and the company is explicit that Opus 5.5 remains clearly stronger on complex, open-ended work that needs sustained judgment. Sonnet 5.5 is positioned for well-scoped everyday tasks, bug fixing and polished documents, slides and spreadsheets.

Why Cost Per Task Is The Number To Watch

Anthropic says Sonnet 5.5 costs up to 30 percent less per task than Sonnet 5 and generates output more than 30 percent faster, making it the fastest Sonnet so far. A flat price list hides that, because the bill depends on how many tokens a job takes.

The effort setting moves the bill further. Anthropic says that at medium effort, the default in the Claude apps, Sonnet 5.5 beats Sonnet 5's best Terminal-Bench score for less than a tenth of the cost per task. At high effort, the default on the Claude Platform, it matches GPT-6 Sol's best FrontierCode score for about a fifth of the cost per task. Anthropic notes that GPT-6 Sol was not reported on some benchmarks, so it used GPT-5.6 Sol there.

The risk sits at the other end. Simon Willison, who tested the model on launch day, reported that at max effort one prompt ran for 128,000 tokens at a cost of $1.28 and failed to produce an answer. His xhigh run of the same prompt cost 5.74 cents and took 41 seconds.

What Else Changed With This Release

Sonnet 5.5 is the first Sonnet model to launch with cyber safeguards and fallbacks like those Anthropic built for its most capable models, because its cybersecurity capabilities are comparable to Opus 5 level. The safeguards target a narrow set of high-risk requests, and Anthropic says routine software development and most life-sciences work are unaffected.

Willison also noted that Sonnet 5.5 is now the model behind the free tier on claude.ai, and said that on his reading ChatGPT's free tier runs the weaker Luna 5.6. If accurate, a free account now reaches a stronger model on Anthropic's side.

Anthropic repeated that Claude Haiku 5.5, built for high-volume and cost-sensitive work, will join the family in the coming weeks.

What To Do Before You Switch

Re-run your own tasks at low and medium effort, not only at the top setting, and compare cost per finished job rather than the token price. Vendor benchmarks do not tell you how many tokens your prompts will burn.

Set a hard token cap and a default effort per workload before the model goes into an agent loop, because the failure Willison hit costs real money when it repeats at scale. If your team does security testing, check how the new safeguards treat your requests.

Keep Opus-class models for the open-ended judgment work Anthropic itself reserves for them, and route the well-scoped volume to Sonnet.

Update, 7 October 2026: Anthropic released Claude Haiku 5.5 as promised and halved the cache-read price of Sonnet 5.5 from $0.20 to $0.10 per million tokens, which it says makes Sonnet around 20 percent cheaper on most agentic work.

Servola Journal

We do this for everyone trying to keep up with what technology is doing to our lives. The people who build it, and the people it happens to. The Servola Journal exists so that what we learn belongs to all of them.

Nobody pays us for this. No ads, no paywall, free to everyone. We just believe that understanding what's happening to all of us shouldn't depend on who can afford to pay for it.

If it gave you something today, tell us to keep going. Follow us, leave a like, or write a positive comment. We read every one, and they are what keeps us going.

Read next: Haiku 5.5 Is 10x Cheaper Per Token, Not Per Job  |  Your Cybersecurity Prompts Just Got Quietly Rerouted