According to CoinMeta, OpenAI launched GPT-6.1 and SOL one week after releasing GPT-6 and SOL, with a focus on improving performance in programming and complex tasks while keeping the price unchanged. GPT-6.1 and SOL achieved 75.2% in the Deepswe 1.1 programming test, which is 6.4 percentage points higher than GPT-6 and SOL, and is now comparable to Astra. In computer usage tests, it lags behind by only 2.1 percentage points. According to the latest artificial intelligence index, GPT-6.1 and SOL scored 52, GPT-6 and Astra scored 53, and GPT-6 and SOL scored 48. The standard price for API is $2 per million token inputs, with an output of $10; the cost for caching inputs has been reduced to $0.1, which is one-fifth of the $10 input and $50 output of Astra. GPT-6.1 and SOL have already been launched, and can be used by API, Codex users, as well as Plus and above; they have not yet entered the general ChatGPT dialogue mode. The initial community feedback focused on cost-effectiveness, with developers believing it to be more suitable as a daily coding assistant, but it is still more stable in complex tasks.