Anthropic launches a faster mid-tier Claude for everyday office work
Reviewed by Helen Jones
On 28 September 2026 Anthropic released Claude Sonnet 5.5, replacing Sonnet 5 as its mid-tier model. It says the new model runs more than 30 percent faster and costs up to 30 percent less for most billed work, even though list prices stay at $2 per million input tokens and $10 per million output tokens. Reuters noted Opus 5.5 arrived the week before at $4 input and $20 output per million tokens. Anthropic is pitching Sonnet 5.5 for well-scoped daily tasks such as documents, slides, spreadsheets, and coding, and as a cheaper partner to Opus. A single Claude chat that no longer asks you to pick a mode first is rolling out gradually to Pro and Max on web, desktop, and mobile, carrying over chats, projects, connectors, and skills. Haiku 5.5 is described as coming soon.
Until this week, many teams treated the mid-tier Claude as last month’s default and jumped to the larger model whenever a report, deck, or analysis felt important. That habit mixed one-off chats with extra checking, because the expensive model still needed the same business context and review steps. The new mid-tier is now the intended workhorse for scoped office tasks, while Opus sits as the premium option. Same sticker price as Sonnet 5 with fewer tokens used to finish the job is a cost and speed change, not a reason to skip quality checks. The unified chat surface only helps if your projects, context, and review rules already live there.
Analysis
Treat this as a default-model test, not a full stack change. Re-run one existing Claude Project or reusable briefing on Sonnet 5.5, compare it with last week’s output for accuracy, tone, and token use, then only switch the team default if the review burden did not grow.
Source note
Pulse published by Collab365 Spaces, reviewed by Helen Jones on . Cite as "Anthropic launches a faster mid-tier Claude for everyday office work", Collab365 Spaces. 4 sources referenced.