Key Notes
- Anthropic released Claude Sonnet 5.5, priced the same as Sonnet 5 but running over 30 percent faster and costing up to 30 percent less per task due to lower token usage.
- The model scored 70.6 percent on Terminal-Bench 4.0 versus Sonnet 5's 10.3 percent, and nearly matched Opus 5.5 on GDPval-AA knowledge work tasks.
- Sonnet 5.5 launches with cyber safeguards similar to Anthropic's most capable models and is the first Sonnet model to include safety classifiers against distillation attacks.
Anthropic released Claude Sonnet 5.5 on September 28, the second model in its Claude 5.5 family, pitching it as a clear upgrade over Sonnet 5 that runs more than 30 percent faster and costs up to 30 percent less for most work, at unchanged pricing.
Sonnet 5.5 is positioned as a faster, lower cost complement to Opus 5.5, Anthropic’s most capable model. Where Opus 5.5 is built for complex work that requires careful judgment, Anthropic says Sonnet 5.5 is strongest at well scoped everyday tasks, fixing bugs, and producing polished documents, slides and spreadsheets.
Sharp Gains in Coding Benchmarks
Anthropic’s own figures show Sonnet 5.5 scoring 70.6 percent on Terminal-Bench 4.0, an agentic coding evaluation, compared with 10.3 percent for Sonnet 5. On CursorBench 4.0, which tests coding agents on ambiguous, multi file tasks drawn from real Cursor sessions, Sonnet 5.5 scored 55.5 percent, within about two points of Opus 5.5’s 57.8 percent and ahead of Sonnet 5’s 34.1 percent.
On FrontierCode 1.1, which measures whether an agent’s code changes would actually be merged, Sonnet 5.5 scored 10 points higher than Sonnet 5 at the same effort setting, at roughly one fifteenth of the cost per task, according to the company.
On GDPval-AA, a test of real world work across 44 occupations and nine industries, Sonnet 5.5 scored nearly level with Opus 5.5 and about 400 points above Sonnet 5. Anthropic also said Sonnet 5.5 is the first Sonnet model to beat the video game Pokemon Red working only from screenshots, a benchmark the company has used previously to illustrate long horizon planning and image understanding.
Pricing Held Flat
Sonnet 5.5 is priced the same as Sonnet 5: 2 dollars per million input tokens, 10 dollars per million output tokens, and 0.20 dollars per million tokens for cache reads. Because the model typically needs fewer tokens to complete the same work, Anthropic says it costs up to 30 percent less per task in practice despite the unchanged headline rate.
Opus 5.5, by comparison, is priced at 4 dollars per million input tokens and 20 dollars per million output tokens.
Safety Changes Tied to Higher Capability
Because Sonnet 5.5’s cybersecurity capabilities are now comparable to Opus 5’s, Anthropic is deploying it with cyber safeguards similar to those used on its most capable models, meaning higher risk cybersecurity tasks will visibly fall back to Sonnet 5 while routine software development is unaffected.
Its biology safeguards remain the same as Sonnet 5’s. Sonnet 5.5 is also the first Sonnet model to launch with safety classifiers designed to prevent distillation attacks, in which attackers use large numbers of fake accounts to extract a model’s capabilities at scale without its built in safeguards.
On Anthropic’s automated behavioral audit, which tests Claude across roughly 1,850 scenarios, the company said Sonnet 5.5 improves on or matches Sonnet 5 on most measures of alignment and honesty, and comes close to Opus 5.5, its best performing model on that measure, in how rarely it tries to escape its sandbox during testing. Anthropic cautioned that no set of evaluations reliably catches every failure mode, and that Sonnet 5.5 may have tendencies the company has not yet identified.
Availability and What Comes Next
Claude Sonnet 5.5 is available now on the Claude Platform, Amazon Web Services, Google Cloud and Microsoft Azure, with zero data retention as an option. Anthropic said Claude Haiku 5.5, a smaller model built for high volume, cost sensitive applications, will join the Claude 5.5 family in the coming weeks, which would complete the lineup Anthropic has been rolling out following Opus 5.5’s release last week.
The release lands as Anthropic works through its own confidential IPO filing, which has separately disclosed steep losses alongside warnings about the risks its models could pose, a tension between rapid product releases and calls from its own chief executive for the industry to slow down and coordinate on safety that the company has not directly addressed in its Sonnet 5.5 announcement.
The pace also keeps pressure on rivals such as OpenAI, whose GPT-6 Astra remains Anthropic’s main point of comparison in its own published benchmark charts.
Disclaimer: AIstify is an independent media brand owned and operated by NuvexMedia LLC, publishing news, research, and insights on artificial intelligence, emerging technologies, automation, and related industries. NuvexMedia LLC invests in and collaborates with companies across the AI, technology, software, and digital innovation sectors. These relationships do not influence AIstify’s editorial coverage, and the publication maintains full editorial independence to provide accurate, timely, and objective information. © 2026 NuvexMedia LLC. All rights reserved. This content is for informational purposes only and should not be considered legal, tax, investment, financial, or other professional advice.
