GPT-6.1 Sol: near-Astra work at a fifth of the price

OpenAI has launched GPT-6.1 Sol, an upgrade to the GPT-6 Sol model the company shipped last week alongside GPT-6 Luna. The pitch: it approaches GPT-6 Astra, its current flagship, on agentic coding, computer use, and professional work, at one-fifth of Astra's standard input and output token prices. Standard API pricing lands at $2 per million input tokens, $10 per million output tokens, and $0.10 per million cached input tokens.

That cached-input number is the one to watch. Astra's cached input costs $1.00 per million tokens, so Sol's $0.10 is a tenth of that and, per OpenAI, 50% below the cached pricing of the previous GPT-6 Sol. If you run agents that reuse context across requests, the spread between standard and cached input is where most of the bill comes from: a tenfold gap on cached pricing is what moves the math on high-volume agent workloads.

Benchmark numbers, with the usual caveats

Everything below comes from OpenAI's own announcement; the company notes that competitor scores were taken from publicly available reports.

• Coding. On DeepSWE v1.1, which tests complex software-engineering tasks in real codebases, GPT-6.1 Sol matches Astra at roughly one-fifth the cost, and beats the previous GPT-6 Sol's best score by 6.4 percentage points at a lower reasoning effort.
• Professional documents. On GDP.pdf, which measures answers to professional questions over complex PDFs with tables, charts, and fine print, it outscores Opus 5.5 with fallbacks at less than half the cost per task, and approaches Astra's state of the art at about one-fifth the cost.
• Business workflows. On AutomationBench, it scores 2.2 points above Opus 5.5 at medium reasoning effort at roughly a third of the cost, and 4.8 points above GPT-6 Sol at the same setting.
• Computer use. On OSWorld 2.0's offline set, it improves on GPT-6 Sol by seven percentage points at maximum reasoning effort for less than half the cost, and lands within 2.1 points of Astra at roughly one-seventh the cost per task.
• Scientific work. On Terminal-Bench Science 0.1, it more than doubles GPT-6 Sol's maximum-effort score at less than half the cost per task: $5.47 per task on average, versus $23.21 for Opus 5.5 and $23.80 for Astra. Astra still tops the board at 68.1%, and OpenAI says it remains the model to use for the hardest scientific tasks.

On factuality, the biggest gain over GPT-6 Sol comes at low reasoning effort, where it cuts the share of responses containing a factual error from 11.4% to 7.7%, about a 32% reduction. Across the tested settings, its error rate stays within 1.9 percentage points of Astra's at less than one-fifth the cost. OpenAI is upfront that these prompts come from de-identified conversations where users flagged an earlier model's error, so they are harder than typical usage.

Safety and availability

The system card addendum reports alignment improvements over GPT-6 Sol, bringing it closer to Astra. Per OpenAI, the model is more transparent about its limitations, more reliable at respecting explicit restrictions, and showed no attempts to bypass its automated safety reviewer, the same behavior as Astra and the previous Sol.

GPT-6.1 Sol is available now in ChatGPT Work and Codex for Plus, Pro, Business, Enterprise, and Edu users, and through the API as gpt-6.1-sol. It is not yet available in Chat. OpenAI says a GPT-6.1 Sol Ultrafast variant, with up to 8x faster token generation in Codex, will follow in the coming days.

What to do with this

For a team deciding where to point its agent traffic, the practical read: if your workload is coding, document-heavy professional work, or computer use, and you are currently on Astra or a comparable frontier model, this is a real step down in cost with a modest capability trade-off, per OpenAI's numbers. If you run the hardest scientific-research tasks, keep Astra for those. And as with any vendor benchmark set, the gaps close or open once you test on your own workloads; the cached-input price is the part worth verifying first, since that is where the savings concentrate.