Anthropic releases Sonnet 5.5 with faster output and unchanged token prices

September 29, 2026

Anthropic released Claude Sonnet 5.5 on September 28, positioning it for everyday coding and document work. The company says output generation is more than 30% faster than Sonnet 5 and that many tasks cost up to 30% less in its testing.

Lower task costs, not lower token rates

Published prices remain $2 per million input tokens and $10 per million output tokens, with cache reads at $0.20 per million tokens. Anthropic attributes the savings to using fewer tokens to complete work. These are vendor-reported results, rather than a guarantee for every application.

Anthropic also reports a 70.6% score on Terminal-Bench 4.0, compared with 10.3% for Sonnet 5. Haiku 5.5 is planned for the coming weeks.

Two AWS routes

AWS confirmed availability through Amazon Bedrock and Claude Platform on AWS. Bedrock offers AWS-managed features such as Guardrails and Knowledge Bases. Claude Platform on AWS provides Anthropic’s native platform experience with AWS billing and authentication.

Sources: Anthropic launch announcement; AWS availability announcement.

Feature image: AI-generated editorial illustration for Guava Insights. Conceptual imagery, not a product photograph or game screenshot.

Topics:
Read next
October 23, 2026

OpenAI Reportedly Seeks $30B at About $1.4T Valuation

Reports from early October say OpenAI is seeking at least $30 billion at about a $1.4 trillion pre-money valuation, with UAE funds up to $10 billion and BlackRock in talks, and an IPO pushed to at least 2027. Nothing is confirmed.
October 23, 2026

Walmart Opens to AI Shopping Agents While Amazon Blocks Them

On October 21, Walmart announced an MCP server that lets approved AI agents search inventory and buy for shoppers at Walmart and Sam’s Club, with spending limits and per-agent profiles. Amazon, meanwhile, is blocking agent purchases.
October 23, 2026

Microsoft MAI-DxO in Science: 83% on Hard Cases, With Caveats

A peer-reviewed Science study reports Microsoft’s MAI-DxO diagnosing 83.2% of 304 complex NEJM cases versus 36.1% for generalist physicians. The setup matters: text-only cases, no exams, no imaging, no clinical deployment yet.