OpenAI introduced GPT-6.1 Sol, a model for coding and computer-use work that comes close to flagship GPT-6 Astra in quality while costing noticeably less per task.

What it costs

Standard API pricing: $2 per million input tokens, $10 per million output tokens, and just $0.10 for cached input. On the DeepSWE v1.1 coding benchmark the model scored 75.2%, up from 68.8% on the previous version in the same line — at roughly 76% lower cost for the same task.

What it means for business

For everyday coding work, you no longer need the most expensive model — the mid-tier option barely loses on quality. The rule is simple: send rare, complex decisions to the top model, and frequent repetitive work to the cheaper one. Over a month, that price gap adds up fast.

We split work across our own Milas the same way: the strong model thinks, the cheaper one handles the routine.

Source