OpenAI has released GPT-6.1 Sol, an upgrade to GPT-6 Sol that comes close to GPT-6 Astra on agentic coding, computer use, and professional work. It costs about one-fifth of what GPT-6 Astra does through the API.
The announcement arrives a week after OpenAI launched GPT-6 Sol and GPT-6 Luna, and a few weeks after GPT-6 Astra became the company’s top model. GPT-6.1 Sol is available starting today in ChatGPT Work and Codex.
GPT-6.1 Sol Costs $2 per Million Input Tokens Against $10 for GPT-6 Astra
OpenAI now sells three GPT-6 tiers, and GPT-6.1 Sol sits in the middle. Standard API prices per million tokens look like this:
- GPT-6 Astra: $10 input, $50 output, $1 cached input
- GPT-6.1 Sol: $2 input, $10 output, $0.10 cached input
- GPT-6 Luna: $0.10 input, $0.50 output, $0.01 cached input
Cached input at $0.10 per million tokens is 95 percent below GPT-6.1 Sol’s own standard input price and 50 percent below what GPT-6 Sol charged for cached input. Agents benefit most, since they resend the same context on every request.
GPT-6.1 Sol Matches GPT-6 Astra on DeepSWE and Beats Claude Opus 5.5 on Professional Benchmarks
On DeepSWE v1.1, a test of complex software engineering tasks in real codebases, GPT-6.1 Sol matches GPT-6 Astra at roughly one-fifth of the cost. It also beats GPT-6 Sol’s best score by 6.4 percentage points while using lower reasoning effort.
The professional and computer-use results are close behind:
- On GDP.pdf, which asks questions about complex PDFs from finance, healthcare, legal, and other fields, GPT-6.1 Sol scores higher than Claude Opus 5.5 with fallbacks at less than half the cost per task. It approaches GPT-6 Astra at about one-fifth the cost.
- On AutomationBench, which tests multi-step business workflows across 47 tools, GPT-6.1 Sol lands 2.2 points above Claude Opus 5.5 at medium effort for roughly a third of the cost. That is also 4.8 points above GPT-6 Sol.
- On OSWorld 2.0, an offline set of long computer-use workflows, GPT-6.1 Sol beats GPT-6 Sol by seven points at maximum effort for less than half the cost. It trails GPT-6 Astra by 2.1 points at about one-seventh the cost per task.
Scientific work is the one area where OpenAI tells users to stay on the bigger model. On Terminal-Bench Science 0.1, GPT-6.1 Sol more than doubles GPT-6 Sol’s score at maximum effort, and it costs an average of $5.47 per task. Claude Opus 5.5 averages $23.21 and GPT-6 Astra $23.80. GPT-6 Astra still posts the top score at 68.1 percent, and OpenAI recommends it for the hardest research tasks.
GPT-6.1 Sol also makes fewer factual errors at low reasoning effort, where the share of responses containing an error falls from 11.4 percent to 7.7 percent, a drop of about 32 percent from GPT-6 Sol. OpenAI ran that test on conversations where users had flagged an earlier model’s mistake, so the prompts are harder than typical use.
OpenAI also tested what happens when a search tool breaks. GPT-6.1 Sol fails to tell the user in 2.1 percent of cases, down from 4.9 percent for GPT-6 Sol. GPT-6 Astra sits at 1.5 percent and GPT-6 Luna at 28.7 percent.
GPT-6.1 Sol Is Live in ChatGPT Work and Codex but Not Yet in Chat
GPT-6.1 Sol is rolling out to Plus, Pro, Business, Enterprise, and Edu users in ChatGPT Work and Codex. It is not available in regular ChatGPT chat yet. Developers can call it in the API as gpt-6.1-sol.
OpenAI also plans to add GPT-6.1 Sol Ultrafast to Codex in the coming days, with token generation up to 8x faster than standard speed. It follows the Ultrafast preview OpenAI showed earlier with GPT-5.6 Sol.
