OpenAI is pricing GPT-6.1 Sol, its upgrade to GPT-6 Sol, at one-fifth of GPT-6 Astra's standard token rates, and says that in its own tests the cheaper model nearly matches Astra on agentic coding, computer use and professional work.
Pricing carries most of the story. Standard API rates are $2 per million input tokens and $10 per million output tokens, per OpenAI, and cached input costs $0.10 per million. That cached rate is 95% below standard input and half of what GPT-6 Sol charged, which matters for agents that resend the same context on every step.
The comparison with Astra is close but not uniform. On DeepSWE v1.1, a software-engineering test built on real codebases, GPT-6.1 Sol matches Astra at roughly a fifth of the cost. On the offline set of the OSWorld computer-use benchmark, it lands 2.1 points behind Astra at maximum effort, for about one-seventh of the cost per task.
Science is the clearest case where Astra stays ahead. It still posts the top score on the Terminal-Bench Science test at 68.1%, and OpenAI itself says Astra should handle the most difficult research tasks. Sol averages $5.47 per task at maximum effort against $23.80 for Astra. On factuality, the share of low-effort Sol answers containing an error fell from 11.4% to 7.7% on de-identified ChatGPT conversations where users had flagged an earlier model's mistake, though OpenAI notes the prompts were chosen to be hard and do not reflect typical use.
The safety results come from tasks selected to elicit failures. In a broken-search-tool test, Sol failed to tell the user in 2.1% of cases, against 4.9% for GPT-6 Sol and 1.5% for Astra. OpenAI says it saw no attempts to bypass an automated safety reviewer, and it points to a system card addendum for the full evaluations.
GPT-6.1 Sol is available to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, though not yet in Chat, and developers can call it in the API as gpt-6.1-sol. A GPT-6.1 Sol Ultrafast option, with up to 8x faster token generation in Codex, is due in the coming days; the announcement lists no price for it. For teams that run agents around the clock, the numbers read as a routing rule: Astra for the hardest jobs, Sol for the volume.
