Anthropic releases Claude Sonnet 5.5 with a massive leap in coding performance

Anthropic just updated its mid-tier AI model, and it runs complex coding tasks at a fraction of the previous cost.

Anthropic released Claude Sonnet 5.5, the second model in its 5.5 family. According to Anthropic, it runs over 30% faster than Sonnet 5 and cuts task costs by up to 30%. Per-token pricing stays the same—$2 per million input tokens and $10 per million output tokens—but the model batches tool calls to complete work in fewer steps.

Why it matters: Sonnet 5.5 closes the gap between mid-tier and flagship performance. On Terminal-Bench 4.0, a command-line coding test, Sonnet 5.5 scored 70.6%, up from Sonnet 5’s 10.3%. Developers get near-flagship performance for writing code, fixing bugs, and polishing interfaces without paying Opus-level API bills.

Here's the gist: at low or medium effort settings, Sonnet 5.5 beats Sonnet 5's best benchmark scores for about a tenth of the cost per task. It is also the first Sonnet model to launch with dedicated cyber safeguards, and the first to beat Pokémon Red using only screenshots.

If you want top-tier software assistance on a everyday API budget, this is your model.