OpenAI rolls out GPT-6 Sol update in just one week, with its performance approaching Astra.

Beating AI Express: Just a week after launching GPT-6 Sol, OpenAI has rolled out GPT-6.1 Sol. The new model focuses on improving performance in programming, Computer Use, and complex professional tasks, with prices remaining largely unchanged. In OpenAI's internal tests, GPT-6.1 Sol scored 75.2% on the DeepSWE 1.1 programming benchmark, 6.4 percentage points higher than GPT-6 Sol's top score, and is now on par with Astra. It also lagged Astra by just 2.1 percentage points in the Computer Use benchmark. Third-party tests yielded similar results. In Artificial Analysis' latest Intelligence Index, GPT-6.1 Sol Max scored 52 points, GPT-6 Astra scored 53, and GPT-6 Sol scored 48. The standard API pricing remains unchanged at $2 per million input tokens and $10 per million output tokens, matching GPT-6 Sol. Cached input pricing has also been reduced from $0.2 to $0.1 per million tokens. In comparison, Astra's standard pricing is $10 per million input tokens and $50 per million output tokens, making GPT-6.1 Sol's standard price just one-fifth of Astra's. GPT-6.1 Sol is now available via API, Codex, and ChatGPT Work, accessible to Plus and higher-tier users, though it has not yet been integrated into standard ChatGPT conversations. Early community feedback has focused heavily on cost-effectiveness. Many developers note it is significantly more suitable than GPT-6 Sol for use as a daily coding agent, though some real-world tests still find Astra more reliable for complex tasks. The move effectively brings most of the flagship model's capabilities down to a price point that supports frequent, high-volume use.