Anthropic released Claude Sonnet 5.5, the second model of the 5.5 family after Opus 5.5

Claude Sonnet 5.5 (@claudeai)

The model runs more than 30% faster than Sonnet 5 and costs up to 30% less per task, while the per token price is unchanged.

For developers: connect to the claude-sonnet-5-5 model on Claude Platform. Those running Sonnet with thinking turned off should first switch to between_tools using the migration guide.

Facts

On September 28, 2026 Anthropic introduced Claude Sonnet 5.5, the second model of the Claude 5.5 family. According to it, this is a clear upgrade over Sonnet 5: more than 30% faster and up to 30% cheaper on most tasks. [@claudeai, September 28, 2026]
This is the first Sonnet with cyber protection and fallbacks like the most powerful models. Anthropic states that normal development is not affected. [@claudeai, September 28, 2026]
Sonnet 5.5 is available everywhere starting today, Haiku 5.5 will come out in the coming weeks. If you are running Sonnet with thinking turned off, before migrating you need to switch to the new between_tools setting. [@claudeai, September 28, 2026]
Anthropic writes that the price is the same as Sonnet 5, but the model needs noticeably fewer tokens for the same work. This is the fastest Sonnet to date. Support for the $2 per million input and $10 per million output token figures comes from a retelling on X, not from Anthropic itself. [@AIScientist_X, September 29, 2026]
On Terminal-Bench 4.0 Sonnet 5.5 scores 70.6% versus 10.3% for Sonnet 5 and 66.4% for Opus 5.5 on xhigh. On CursorBench 4.0 Sonnet 5.5 scores 55.5%, Opus 5.5 scores 57.8%. The figures are its own, from the Anthropic release page. [anthropic.com]

Waiting for

Exactly when Claude Haiku 5.5 will come out: Anthropic says only "in the coming weeks".

post from telegram channel trad_fin · post from telegram channel Rickler Feed · post from telegram channel ahboyashreads