OpenAI released GPT-6.1 Sol on September 29, 2026 as a model aimed at complex coding, computer use, and professional workloads. OpenAI positions it as a lower-cost option for demanding tasks, with support for tool use and multi-agent workflows.
Model capabilities
GPT-6.1 Sol is available through the Responses API and Chat Completions, with tool calling primarily centered on the Responses API. The model supports reasoning, structured outputs, web search, file search, code interpreter, hosted shell, computer use, MCP, and other agent-oriented tools.
The model has a 1,050,000-token context window and supports up to 128,000 output tokens. OpenAI lists reasoning effort settings from low through max, giving developers more control over the balance between latency and depth.
Pricing
Standard pricing for prompts with up to 272K input tokens is $2 per million input tokens and $10 per million output tokens. Cached input is $0.10 per million tokens, while cache writes are $2.50 per million. Larger prompts above 272K input tokens use different rates.
Multi-agent support
GPT-6.1 Sol also supports multi-agent use in beta. An application can allow the model to delegate parts of a task to subagents within a Responses API request. This can be useful for complex research, software engineering, and workflows where different subtasks can be handled independently.
Where it fits
The model is intended for work where a simple fast model may not provide enough reasoning depth. Coding agents, browser automation, professional research, document processing, and tool-heavy workflows are natural areas to test.
Practical takeaway: Developers should compare GPT-6.1 Sol with their current model using real workloads. The most useful metrics are successful task completion, cost per completed task, latency, tool reliability, and the amount of human correction required.