OpenAI has brought additional capacity online after demand for GPT-6.1 Sol put ChatGPT and Codex under heavy load. Speed should improve over the coming hours to almost twice the previous day's level, OpenAI's Tibo Sottiaux said on Oct. 1.

Sottiaux described GPT-6.1 Sol as “our most demanded model pretty much ever” across both the API and subscriptions. Earlier testing by Vals AI found it ran two to three times slower than GPT-6 Sol on tasks performed by AI agents, taking about 75 minutes per Vibe Code Bench task and 43 minutes per Legal Agent task.

Launched Sept. 29, GPT-6.1 Sol costs $2 per million input tokens and $10 per million output tokens, one-fifth of Astra's standard prices. It scored one point below GPT-6 Astra on the Artificial Analysis Intelligence Index, while costing $0.72 per task at maximum reasoning effort versus Astra's $3.26. Cached input costs $0.10 per million tokens, half the price for GPT-6 Sol.