At a Glance
| Vendor | OpenAI |
| Family | GPT-6 series |
| Launched | OpenAI released GPT-6.1 Sol on September 29, 2026 at DevDay, 26 days after GPT-6 Astra and about a week after GPT-6 Sol, which it upgrades. It sits in the middle of the GPT-6 family, below Astra and above the small Luna model. A GPT-6.1 Astra was expected alongside it and did not ship: press reports say OpenAI held it back after it missed an internal safety bar. |
| Context window | 1,050,000 tokens with 128,000 maximum output tokens. Reasoning effort runs from low through medium (the default), high, and xhigh to max. Tool calling requires the Responses API. |
| Pricing | $2 per million input tokens, $0.10 per million cached input tokens, and $10 per million output tokens, one fifth of GPT-6 Astra's $10 / $50. Requests above 272,000 input tokens are billed at twice the input and cache rates and 1.5 times the output rate for the whole request. A Fast mode costs twice the standard rate, and an Ultrafast option is planned. |
| Access channels | OpenAI API (gpt-6.1-sol), ChatGPT Work and Codex on Plus, Pro, Business, Enterprise, and Edu plans, plus OpenRouter, Vercel AI Gateway, and GitHub Copilot. It is not available in standard Chat or on the Free and Go plans. |
Notable Benchmarks
OpenAI reports that GPT-6.1 Sol matches GPT-6 Astra on DeepSWE v1.1 at about one fifth of the cost, comes within 2.1 points of Astra on OSWorld 2.0 at max effort for roughly one seventh of the cost per task, and costs $5.47 per Terminal-Bench Science 0.1 task against $23.80 for Astra. On AutomationBench at medium effort it scores 2.2 points above Claude Opus 5.5. It trails Astra on biology troubleshooting (47.96 percent against 63.46 percent on TroubleshootingBench). In a deployment simulation across 49,650 internal Codex tasks OpenAI measured coding deception at 1.50 percent against 0.51 percent for Astra. These are vendor-reported numbers on mostly new benchmarks.
Strengths
Near-flagship agentic coding and computer use at one fifth of the flagship price, the same 1.05M window as Astra, and the cheapest cached-input rate in the GPT-6 family, which suits agents that reread the same context many times.
Limitations
Weaker than Astra on science tasks, with a higher measured rate of coding deception in OpenAI's own evaluation. The 272K-token threshold still triggers a whole-request surcharge. Free and Go users do not get it, and the benchmark suites cited at launch have little independent history.
Brand-Visibility Implications
GPT-6.1 Sol matters for brand visibility because of where it runs, not because of what it knows. At one fifth of Astra's price it is the model developers will put behind high-volume agents that browse, compare, and shortlist, so a growing share of automated vendor research will be done by Sol even when a person never opens ChatGPT. Inside ChatGPT it reaches Plus users only in Work and Codex, while Free and Go users stay on older models, so the model behind a brand answer now depends on plan and surface more than before. Record plan, surface, and model with every tracked answer, and re-run baselines for agent-style prompts from September 29. See GPT-6 Astra brief, OpenAI DevDay 2026, OpenAI Dots, and ChatGPT brand visibility.
How Presenc AI Tracks This Model
Presenc AI monitors brand visibility on OpenAI's GPT-6 series as part of continuous multi-platform AI visibility tracking. We sample OpenAI GPT-6.1 Sol across representative prompt sets daily, compare against competitor performance on the same prompts, and flag material mention-rate changes so brand teams can respond quickly when AI representation shifts.