OpenAI has introduced the all-new GPT-6.1 Sol. The newly launched model is placed between the GPT-6 Astra and GPT-6 Luna. According to the AI giant, the GPT-6.1 Sol could come close to GPT-6 Astra on several demanding tasks, including coding, professional work and computer use. However, at the same time, the API pricing of the new model is set at one-fifth of Astra’s standard input and output token prices. Follow along with the article to learn more about the all-new GPT-6.1 Sol model.
According to the data shared by OpenAI in its blog post, the all-new GPT-6.1 Sol is much better than the earlier GPT-6 Sol, especially at handling difficult tasks such as coding and debugging, understanding documents, and completing tasks that involve several steps. Not only that, but the company also claims that in some of these areas, the GPT-6.1 Sol performs almost as well as GPT-6 Astra. OpenAI shared that in the DeepSWE 1.1 coding benchmark, the GPT-6.1 Sol matched GPT-6 Astra while costing about one-fifth as much per task.
GPT-6.1 Sol is here.
— OpenAI Developers (@OpenAIDevs) September 29, 2026
Upgraded with stronger agentic coding and computer use, near-Astra performance, and cached input at a 95% discount to standard input pricing.
GPT-6.1 Sol is built for complex refactors, deep codebase investigations, and long-running agents across apps. pic.twitter.com/1XRVKHgISH
Furthermore, the new model is also said to be more accurate when dealing with difficult questions. An AI giant claims that its biggest improvement over GPT-6 Sol comes when using low reasoning effort. In this setting, factual mistakes fall from 11.4 per cent to 7.7 per cent. OpenAI also claims that across the tested reasoning settings, GPT-6.1 Sol’s error rate is within 1.9 percentage points of GPT-6 Astra. These tests use deliberately difficult prompts and are not representative of typical usage.
Also read: OpenAI dots announced at DevDay 2026: How these always-on AI agents will handle tasks on your behalf
OpenAI also says GPT-6.1 Sol is clearer about what it cannot do and does a better job of following what users ask while respecting safety rules. When put in challenging situations like flagging broken search tools, following explicit restrictions, and avoiding unauthorised outcomes during tasks, the new model is said to fail less often when compared to GPT-6 Sol. Furthermore, OpenAI also stated that it found no cases where the model tried to bypass the automated safety checks, similar to GPT-6 Astra and GPT-6 Sol.
GPT-6.1 Sol is available starting today to all Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex. Developers can also access the model through the OpenAI API as gpt-6.1-sol. However, users will not be able to access the new model in the chat yet.
Also read: New rules from October 1: Home appliance price hike, UPI MDR changes, ITR deadline updates and more
OpenAI has also pointed out that the GPT-6.1 Sol Ultrafast version will also arrive in the coming days, offering up to eight times faster token generation in Codex.
OpenAI has not yet announced a separate India price for GPT-6.1 Sol. However, the API is priced globally at $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens.