Anthropic has released Claude Sonnet 5.5, a new model priced identically to Sonnet 5 but claimed to be faster and cheaper for most tasks. Benchmarks suggest it outperforms the previous version, yet it still suffers from the same bug found in Opus 5.5 where maximum thinking effort causes the model to exceed token limits and fail.
The most significant change is that Sonnet 5.5 now powers the free tier on claude.ai. This puts Anthropic ahead of OpenAI, which currently relies on Luna 5.6 for its free ChatGPT access, giving users a more capable entry point without payment. The model also performs well on complex coding tasks, including viral 3D animation tricks.
* Available immediately on the paid tiers
* Replaces Sonnet 5 as the default free model
* Haiku 5.5 is expected in coming weeks
