Anthropic is preparing for the release of the Claude Sonnet 5.5 model, which is currently undergoing beta testing within Claude Code. The model has been identified in configuration files alongside other versions, including Claude Opus 5.5 and Claude Fable 5.1. Some users have reported that their accounts have already been routed to the new version for testing.
Early benchmarks suggest that Sonnet 5.5 outperforms OpenAI's GPT-6 Sol in coding tasks and UI animation generation. Reports indicate that the model's performance in advanced coding and agent capabilities is approaching that of the flagship GPT-6 Astra. Developers have noted that the model maintains high efficiency and natural output quality.
The pricing for the new model is set at $2 per million input tokens, with cache reads priced at $0.20 per million tokens. This pricing strategy aims to compete directly with OpenAI's mid-range models. Analysts suggest that Anthropic is leveraging an internal AI-driven development cycle, where 26% of core research is currently conducted by previous Claude models.
The official launch is rumored to coincide with OpenAI's DevDay on September 29. Anthropic previously announced in September that Sonnet 5.5 and Haiku 5.5 would be released within weeks. The company continues to expand its ecosystem, recently launching a marketplace featuring over 2,000 plugins.
