GPT 6.1 Sol: Near-Astra intelligence for a fifth of the price
- Source
- openai.com
- Author
- crorella
- Date
OpenAI positions GPT-6.1 Sol as near-Astra intelligence at roughly a fifth of the price. The list price is $2 per million input tokens and $10 per million output tokens, with cached input discounted 95% to $0.10 per million. OpenAI's developer account reports stronger agentic coding and computer use, and near-Astra results on DeepSWE, AutomationBench and OSWorld 2.0, including 75.2% on DeepSWE v1.1 at high effort, ahead of GPT-6 Sol's best at about 76% lower cost per task.
The independent numbers point the same way. Artificial Analysis places GPT-6.1 Sol one point below GPT-6 Astra on its Intelligence Index, at $0.72 per task at max effort against $3.26 for Astra. It also measured a 12 point gain on Terminal-Bench 4.0 and a lower hallucination rate. Vercel has already made the model callable through AI Gateway.
The practical consequence is a routing decision. Workloads that reuse long shared context benefit twice, from the lower base price and from the deep cache discount. Teams defaulting to Astra should test Sol on their own tasks first and reserve Astra for the cases where the one-point gap shows up in results.
Context
OpenAI positions GPT-6.1 Sol as a lower-cost alternative to its larger GPT-6 Astra model. Its release post is titled "Near-Astra intelligence for a fifth of the price," so the comparison is OpenAI's own and concerns standard rates: $2 per million input and $10 per million output tokens for Sol, which OpenAI frames as one-fifth of Astra's. The closeness in capability is likewise OpenAI's characterization, not an independent measurement, and total task cost also depends on how much a job uses. Cached input is listed at $0.10 per million tokens, so repeated use of an eligible shared costs substantially less than sending it as uncached input.
- token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
- context window — The maximum amount of text a model can consider at once — its working memory for the current conversation or task.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Checking sign-in…
Loading comments…