Claude Sonnet 5.5: Faster Coding & Visual QA

Anthropic pushed out Sonnet 5.5 as the new default model inside the Claude app, and unlike Opus 5.5, it doesn't sit behind a paywall. Anyone using Claude without a subscription is now talking to this model, which puts it in front of far more people than any Opus release will reach. That alone makes it worth a close look, separate from the usual benchmark chart comparisons Anthropic likes to publish.

The pitch here is speed and cost, not raw ceiling performance. Sonnet has always been the middle child of the Claude lineup, and 5.5 leans into that role harder: quick responses, cheap token usage, and agentic behavior tuned for tasks that are well-scoped and repeatable rather than open-ended. Anthropic even ships it with different default effort settings depending on where you access it, Medium in the consumer apps, High on the platform, which changes latency and how much verification the model does on its own output.

What follows is a hands-on look at whether the agentic coding claims hold up, what the benchmarks actually show, and what a free upgrade like this means for people who never touch Opus at all.