OpenAI Challenges Anthropic with GPT-5.6 Model Suite

The flagship model, Sol, is positioned as a direct competitor to Anthropic’s Fable. According to the Artificial Analysis Coding Agent Index, Sol reaches a score of 80, surpassing Fable 5 by 2.8 points while consuming less than half the output tokens. CEO Sam Altman emphasizes that these gains translate to a 54% increase in token efficiency for coding tasks, alongside reduced operational costs. The suite also includes Terra for intermediate needs and the budget-friendly Luna.

Beyond performance metrics, OpenAI is integrating these models into a new tool called ChatGPT Work, designed for enterprise document and spreadsheet management. The company is aggressively targeting the corporate sector to regain market share from Anthropic. Pricing for the new API access is set at $5 input and $30 output for Sol, $2.50 input and $15 output for Terra, and $1 input and $6 output for Luna. The models are available now across ChatGPT, Codex, and the OpenAI API.