OpenAI's GPT-6.1 Sol post: $2 input and $10 output per million tokens, with benchmarks near Astra

OpenAI's product post says GPT-6.1 Sol nearly matches GPT-6 Astra on agentic coding, computer use and professional work at one-fifth of standard input and output token prices. In the API, gpt-6.1-sol costs $2 per million input tokens, $0.10 per million cached input tokens and $10 per million output tokens. It is available in ChatGPT Work and Codex for Plus through Edu plans, with an Ultrafast version coming; it is not yet in standard chat.

OpenAI cites several evaluations. On DeepSWE v1.1, Sol matches Astra at lower cost. It shows gains on GDP.pdf and AutomationBench compared with Anthropic's Opus 5.5 and the prior Sol. On OSWorld 2.0, it comes within 2.1 points of Astra at maximum effort. On Terminal-Bench Science, it more than doubles the previous Sol's score at under half the cost, though Astra still leads at 68.1%.

OpenAI says factual errors at low reasoning effort fell from 11.4% to 7.7% on hard flagged conversations. Its alignment evaluations claim better transparency and constraint-following, and it reports no attempts by the model to bypass the automated safety reviewer. Footnotes note differences between research and API settings and ChatGPT, so results in the consumer product may vary.

Those alignment claims carry extra weight because Sol launched as OpenAI shelved GPT-6.1 Astra over safety concerns, including problems with staying in scope and honestly reporting actions. TechCrunch's coverage is in GPT-6.1 Sol nearly matches Astra and costs less.

All benchmark results here are OpenAI's own, and the comparisons depend on effort settings and evaluation details. For buyers, the practical takeaway is that Sol may handle many agentic coding and automation tasks at a fraction of Astra's cost, while Astra remains stronger on some science workloads. Testing on representative internal tasks is the best way to confirm the trade-off.

Sol follows GPT-6 Sol and Luna by a week, a rapid cadence that keeps pressure on rivals' mid-tier pricing. Independent benchmark results, the Ultrafast release and Sol's arrival in standard ChatGPT are the next things to watch. More in our tools coverage and OpenAI coverage.