Audience
Software developers, engineers, researchers, enterprises, AI application builders, professional knowledge workers, and agent developers that need advanced coding, reasoning, computer-use, scientific, document-analysis, and business-automation capabilities
About GPT-6.1 Sol
GPT-6.1 Sol is an OpenAI model designed to deliver near-GPT-6 Astra intelligence at substantially lower cost across coding, professional work, computer use, scientific research, and agentic workflows. It improves on GPT-6 Sol across complex tasks such as software engineering, document analysis, multi-step business automation, and computer interaction. On DeepSWE v1.1, GPT-6.1 Sol matches GPT-6 Astra at roughly one-fifth of the cost while exceeding GPT-6 Sol's best score by 6.4 percentage points at a lower reasoning effort. The model also improves computer use, outperforming GPT-6 Sol by seven percentage points on the OSWorld 2.0 offline set at maximum reasoning effort while costing less than half as much per task. GPT-6.1 Sol provides improved factuality and alignment, including greater transparency about limitations and stronger adherence to user instructions during agentic tasks. It is available through ChatGPT Work, Codex, and the OpenAI API.
Pricing
Output: $10 per 1 million tokens
Cached Input: $0.10 per 1 million cached input tokens
Company Information
Product Details
GPT-6.1 Sol Frequently Asked Questions
GPT-6.1 Sol Product Features
GPT-6.1 Sol Verified User Reviews
Write a Review-
Probability You Would Recommend?1 2 3 4 5 6 7 8 9 10
"Compares right up to Opus 5.5 and Astra" Posted 2026-10-01
Pros: The biggest improvement for me is that it gets very close to Astra-level performance without Astra-level pricing. I can use it for serious coding, research, computer-use tasks, and long agent workflows without feeling like every run needs to be reserved for something mission-critical. The 1.05M-token context window is still a huge advantage. I can keep large repos, specs, logs, documentation, and a lot of prior agent work in context without constantly trimming things down. I also really like the new multi-agent support. Being able to let Sol delegate parts of a bigger task to subagents makes it much more useful for complicated projects where research, coding, testing, and analysis can happen in parallel.
Cons: The main downside is that Astra is still the model I would reach for when I absolutely want maximum capability and cost is secondary. Sol also no longer supports the none or minimal reasoning settings, so it is less suited to truly lightweight work than Luna.
Overall: Overall, GPT-6.1 Sol feels like the model I would use most often for serious work. It has the context, tools, reasoning, and agent capabilities I want, but the price is low enough that I can actually run it heavily instead of treating it like a special-occasion model.
Read More...
- Previous
- You're on page 1
- Next