OpenAI Launches GPT-6.1 Sol: A Lower-Cost Alternative to GPT-6 Astra for Developer, Business and Professional Use Cases

OpenAI has introduced GPT-6.1 Sol, a new AI model targeted at developers, businesses and professionals seeking advanced reasoning, coding and computer-use capabilities at a lower cost than its flagship GPT-6 Astra model.

According to OpenAI, GPT-6.1 Sol offers performance approaching that of GPT-6 Astra on several challenging tasks, and has a standard API pricing that is lower than the flagship product, with costs of $2 per million input tokens and $10 (per 1 million output tokens) compared to $10 and $50 respectively.

GPT-6.1 Sol is focused on coding and professional AI capabilities

The new model is designed for complicated software engineering, professional, computer-use related tasks and scientific research.

OpenAI states that GPT-6.1 Sol can match GPT-6 Astra on the DeepSWE v1.1 software-engineering benchmark at about one-fifth of the cost per unit. The company also reports improvements over the previous GPT-6 Sol model at a decreased reasoning effort on certain tests.

In addition, the model is designed to handle complicated documents and professional knowledge work. On GDP.pdf, a benchmark that concentrates on answering questions from challenging professional documents, GPT-6.1 Sol reportedly exceeded Opus 5.5 with fallbacks while having a cost advantage per task, according to the company's reported testing results.

Better performance in computer-use and automation

Computer-use capabilities are another strong focus of GPT-6.1 Sol.

OpenAI claims that the model significantly improved on the OSWorld 2.0 benchmark, which measures how AI systems interact with computers and complete tasks. At maximum reasoning effort, GPT-6.1 Sol scored 7 percentage points higher than GPT-6 Sol and within 2.1 percentage points of GPT-6 Astra, according to the company.

Furthermore, the model exhibited improvements on multi-step business workflows. OpenAI reported that GPT-6.1 Sol registered a score that was 2.2 percentage points higher than Opus 5.5 at medium reasoning effort on AutomationBench, and at about one-third the cost per task for the evaluated tasks.

GPT-6.1 Sol is aimed at scientific research

OpenAI is also positioning GPT-6.1 Sol as a model for scientific and technical work.

Based on the company's benchmark results, GPT-6.1 Sol more than doubled GPT-6 Sol's score on Terminal-Bench Science 0.1 at maximum reasoning effort. OpenAI cited an average cost of $5.47 per task for GPT-6.1 Sol compared with $23.80 for GPT-6 Astra on that evaluation.

Moreover, the company reported a decrease in factual errors on challenging prompts, from 11.4% for GPT-6 Sol to 7.7% for GPT-6.1 Sol in its testing. Those numbers are the company's reported evaluation results, rather than an independent assessment.

• GPT-6.1 Sol API Pricing
• GPT-6.1 Sol is priced at:
• Input: $2 per 1 million tokens
• Cached input: $0.10 per 1 million tokens
• Output: $10 per 1 million tokens

OpenAI's pricing documentation also includes batch and flex pricing options while longer-context usage can have different rates.

The lower pricing makes GPT-6.1 Sol especially relevant for applications that process substantial amounts of AI-created or AI-analyzed content.

GPT-6.1 Sol Availability

OpenAI announces that GPT-6.1 Sol is available through the API using the model name gpt-6.1-sol, and it offers capabilities, such as reasoning, streaming, function calling, web search, file search, code interpreter and computer use.

Furthermore, it is being rolled out to several paid ChatGPT plans through ChatGPT Work and Codex. The standard ChatGPT interface availability is different to its API availability.

OpenAI also announced an upcoming GPT-6.1 Sol Ultrafast option for Codex, which is intended to offer significantly faster token generation than the standard version.

What GPT-6.1 Sol Means for AI Developers

GPT-6.1 Sol represents OpenAI's attempt to bring advanced AI capabilities closer to the cost level necessary for large-scale production applications.

Rather than focusing solely on maximizing model intelligence, the new model combines high-end reasoning, coding, computer-use and professional-work capabilities while using substantially lower API pricing. For developers building AI agents, coding assistants and business automation tools, the cost-performance balance is therefore a key consideration.

OpenAI's published evaluations suggest that GPT-6.1 Sol can approach GPT-6 Astra on several complex tasks, although actual performance and cost will vary based on the workload, reasoning settings and application design.