GPT-6.1 Sol promises near-Astra performance at a fraction of the cost

OpenAI has introduced GPT-6.1 Sol, a new AI model positioned between its high-end GPT-6 Astra and lower-cost GPT-6 Luna models. OpenAI announces in an official blog post that Sol is designed for coding, computer use, document analysis, and multi-step business workflows.

The company prices GPT-6.1 Sol at $2 per million input tokens and $10 per million output tokens. Cached input, which applies when developers reuse information from earlier requests, costs $0.10 per million tokens. OpenAI says this is 95% below Sol’s standard input price and half the cached-input price of the earlier GPT-6 Sol.

GPT-6 Astra remains OpenAI’s most capable option, according to the company, but costs five times more for standard input and output. Astra is priced at $10 per million input tokens and $50 per million output tokens. GPT-6 Luna, aimed at high-volume everyday tasks, costs $0.10 for input and $0.50 for output per million tokens.

Performance claims across work tasks

OpenAI reports that GPT-6.1 Sol matches Astra on the DeepSWE 1.1 software engineering benchmark while exceeding the previous Sol model’s best result by 6.4 percentage points. The company also says the model comes close to Astra’s results on complex PDF questions and computer-use tasks.

In Terminal-Bench Science 0.1, a benchmark for research workflows using code and terminal tools, OpenAI says GPT-6.1 Sol costs an average of $5.47 per task at maximum reasoning effort. That compares with $23.80 for Astra. Astra still achieved the highest score in the company’s comparison.

OpenAI also reports improved factual accuracy and safety behavior. At low reasoning effort, the share of responses containing a factual error fell from 11.4% with GPT-6 Sol to 7.7% with the new model, according to the company.

GPT-6.1 Sol is available in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, as well as through the API under the name gpt-6.1-sol. OpenAI says an Ultrafast version will follow in Codex.