What did Anthropic announce on September 28, 2026?

Anthropic released Claude Sonnet 5.5, the second model in its Claude 5.5 family after Opus 5.5, positioning it as a faster, lower-cost model for everyday work including coding and office documents, according to Anthropic's own release notes and TechCrunch, both published September 28, 2026.

Does Sonnet 5.5 cost more or less than Sonnet 5?

Per-token API pricing is unchanged at $2 per million input tokens and $10 per million output tokens, but Anthropic states real-world tasks cost up to 30% less because Sonnet 5.5 needs fewer tokens to reach the same result.

Is Sonnet 5.5 better than Opus 5.5?

Not across the board. Anthropic's own benchmarks show Sonnet 5.5 outperforming Opus 5.5 on agentic coding tasks, while nearly matching it on general knowledge work, but Opus 5.5 remains the higher tier for tasks where Anthropic recommends its strongest reasoning.

News · Published September 30, 2026

Anthropic's Claude Sonnet 5.5: What Actually Changed

Anthropic released Claude Sonnet 5.5 on September 28, 2026, describing it as a faster, cheaper mid-tier model built for everyday work, according to Anthropic's own release notes and TechCrunch's coverage that same day. If you already use Sonnet for coding, drafting documents, or day-to-day chat, this is the update that quietly makes those tasks run faster and burn fewer tokens, without you having to change how you prompt it.

What Anthropic released

Sonnet 5.5 is the second model in the Claude 5.5 family, following Opus 5.5, which Anthropic launched on September 22, 2026. Per Anthropic's release notes, the company calls Sonnet 5.5 "an ideal assistant for everyday tasks, including coding and creating office documents," positioning it below Opus 5.5 on raw capability but ahead of it on cost and speed for routine work. TechCrunch's September 28, 2026 report notes Anthropic plans a new version of Haiku, its smallest model, in the coming weeks, though no release date has been given yet.

How much faster, how much cheaper

According to Anthropic's release notes, Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and needs meaningfully fewer tokens to complete the same task. Per-token API pricing stays the same as Sonnet 5, at $2 per million input tokens and $10 per million output tokens, but Anthropic states that because the model uses fewer tokens per task, real-world costs run up to 30% lower. Box, cited on Anthropic's release page, reported Sonnet 5.5 was "more accurate, 2.4x faster, and used 12% fewer total tokens" in its own testing.

What the benchmarks show

Anthropic's release notes put Sonnet 5.5 at 70.6% on Terminal-Bench 4.0, a jump from Sonnet 5's 10.3% on the same agentic coding benchmark, and 55.5% on CursorBench 4.0. On GDPval-AA, a knowledge-work benchmark, Sonnet 5.5 scored 1844 against Opus 5.5's 1846, effectively even. SpaceXAI, quoted on the same release page, called that CursorBench result "frontier-level performance, second only to Opus 5.5." If you're not sure which Claude model actually fits your work, our guide to choosing a Claude model breaks down Haiku, Sonnet, Opus, and Fable in plain terms.

New to Claude entirely?

Claude AI for Beginners Bible covers the basics this kind of model update builds on, rated 4.5 out of 5 from 303 ratings as of August 30, 2026.

Get the book on Amazon

As an Amazon Associate, this site earns from qualifying purchases.

Where you can use it

Anthropic's release notes list Sonnet 5.5 as available in the Claude apps, the Claude Platform API, and through Amazon Web Services, Google Cloud, and Microsoft Azure, under the model identifier claude-sonnet-5-5. TechCrunch's report frames the release against a busy few weeks for the field, noting recent model updates from OpenAI and a new AI agent model from Meta aimed at smart glasses, without giving Anthropic's rivals equivalent benchmark detail.

What this doesn't tell you yet

This is based on Anthropic's own release notes and TechCrunch's September 28, 2026 report, not independent, hands-on testing of Sonnet 5.5 across everyday tasks. Benchmark scores like Terminal-Bench and CursorBench measure specific coding scenarios, not general writing, research, or conversational use, so your own results may differ. Anthropic did not give an exact date for the promised Haiku update at the time of this reporting.

Questions people are asking about this update

What did Anthropic announce on September 28, 2026?: Anthropic released Claude Sonnet 5.5, the second model in its Claude 5.5 family after Opus 5.5, positioning it as a faster, lower-cost model for everyday work including coding and office documents, according to Anthropic's own release notes and TechCrunch, both published September 28, 2026.
Is Claude Sonnet 5.5 faster than Sonnet 5?: Yes. Per Anthropic's release notes, Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and needs significantly fewer tokens to finish the same work.
Does Sonnet 5.5 cost more or less than Sonnet 5?: Per-token API pricing is unchanged at $2 per million input tokens and $10 per million output tokens, but Anthropic states real-world tasks cost up to 30% less because Sonnet 5.5 needs fewer tokens to reach the same result.
Where can I use Claude Sonnet 5.5?: Anthropic lists it as available in the Claude apps, the Claude Platform API, and through Amazon Web Services, Google Cloud, and Microsoft Azure, under the model identifier claude-sonnet-5-5.
Is Sonnet 5.5 better than Opus 5.5?: Not across the board. Anthropic's own benchmarks show Sonnet 5.5 outperforming Opus 5.5 on agentic coding tasks, while nearly matching it on general knowledge work, but Opus 5.5 remains the higher tier for tasks where Anthropic recommends its strongest reasoning.

Keep reading

Disclosure: this page is an independent reader resource. Some links are Amazon Associates links, which means we may earn a commission from qualifying purchases at no extra cost to you.