Claude Sonnet 5.5 Begins Gray Testing, Performance Rivals GPT-6 Astra
Ma talks about AI2026-9-29

After Anthropic released Opus 5.5, the dedicated configuration identifier for Claude Sonnet 5.5, “claude-sonnet-5-5”, has appeared in the official backend and is now undergoing a small-scale gray release through the Claude Code channel. Internal beta developers report that this new model outperforms GPT-6 Sol in pure frontend UI animation generation and everyday coding scenarios, while its advanced coding and agentic capabilities are already approaching those of GPT-6 Astra. It is priced at $2 per million input tokens, $10 per million output tokens, and only $0.20 per million tokens for cached reads.

Behind this is Anthropic’s competitive move aimed at OpenAI’s annual developer conference: technically matching the performance of top flagship models with lower inference costs, directly capturing users in cost-effective coding and development scenarios. This step almost squeezes out competitors at the same price point.


POPULAR SERVICE PROVIDERS
10 years of cross-border appeal experience; high-difficulty appeals for stores and funds
One-stop service for overseas postcards
Specializing in end-to-end cross-border logistics for Europe, the US, and Canada
Amazon Big Data Product Selection and Operation Software
Wangchen Escort: One-stop cross-border compliance solutions to unlock global business opportunities.