OpenAI launches GPT-6.1 Sol, nearly as powerful as Astra but at one-fifth the price

OpenAI officially launched the GPT-6.1 Sol model at its developer event. The company calls it the most cost-effective model at its performance level, nearly matching the flagship GPT-6 Astra while costing only one-fifth of Astra's standard input and output token price.

GPT-6.1 Sol performs on par with Astra in agentic coding, computer use, and professional workflows. In software engineering tests, it scores 6.4 percentage points higher than GPT-6 Sol at lower reasoning effort. In computer use tests, it scores 7 points higher than GPT-6 Sol at less than half the cost. On science terminal tasks, it scores more than double GPT-6 Sol, with an average cost of $5.47 per task, far below Astra's $23.80.

For reliability, the factual error rate at low reasoning effort dropped from 11.4% with GPT-6 Sol to 7.7%, a roughly 32% decrease. The model better follows user intent and safety constraints, with no attempts to bypass safety reviewers observed in testing.

The model is available to ChatGPT Work and Codex paid users starting September 29, and developers can access it via API as gpt-6.1-sol.