GPT 6.1 Sol Brings Astra-Level Performance at 5x Lower Cost

OpenAI has unveiled GPT-6.1 Sol, a new model that matches the performance of GPT-6 Astra in tasks such as coding, computer use, and professional work, but at a fraction of the cost. This model is an upgrade from GPT-6 Sol, which was released on September 22. Users with Plus, Pro, Business, Enterprise, and Edu subscriptions can access GPT-6.1 Sol via ChatGPT Work and Codex, but it is not yet available in regular ChatGPT conversations.

Developers can integrate the model through the API gpt-6.1-sol. Pricing is significantly lower than GPT-6 Astra, with $2 charged per million input tokens and $10 per million output tokens, compared to Astra's $10 and $50 respectively. The cost for cached input is even lower at $0.10 per million tokens, which is a 95% reduction from the standard input price and half the price of cached input for GPT-6 Sol.

Additionally, OpenAI is planning an Ultrafast version of GPT-6.1 Sol that can generate tokens up to eight times faster in Codex. In benchmark tests, GPT-6.1 Sol performed at or above the level of Astra in various tasks while being 5 times cheaper. For instance, in the DeepSWE v1.1 test, GPT-6.1 Sol scored 6.4 points above GPT-6 Sol and was just 2.1 points behind Astra at roughly one-seventh of the cost.

In AutomationBench, GPT-6.1 Sol outperformed Opus 5.5 by 2 points at about one-third the cost. In OSWorld 2.0, it was within 2.1 points of Astra at roughly one-seventh the cost. GPT-6.1 Sol also demonstrated improved safety features, with lower error rates and fewer instances of bypassing safety restrictions compared to GPT-6 Sol and Astra.