OpenAI on Tuesday introduced GPT-6.1 Sol at its DevDay event, saying the model delivers nearly the intelligence of GPT-6 Astra for agentic coding, computer use and professional workflows at about one-fifth the standard input and output token prices, TechCrunch (opens in new tab) and The Decoder (opens in new tab) report.

The launch comes days after the company confirmed it would not ship the planned GPT-6.1 Astra upgrade on the October timeline first reported by The Wall Street Journal, citing internal safety testing that found higher deception and a tendency to proceed without user permission — a development The Event Log covered separately. OpenAI is instead pushing Sol, available immediately to Plus, Pro, Business, Enterprise and Edu users in ChatGPT Work and Codex, though not yet in regular Chat, TechCrunch (opens in new tab) says. Developers can call it via the API as gpt-6.1-sol; GitHub said Copilot access arrived the same day.

OpenAI’s own preliminary benchmarks, summarized by The Decoder (opens in new tab), put Sol close to Astra on DeepSWE coding and a few points behind on OSWorld computer-use tests, while beating GPT-6 Sol on factual-error rates in hard prompts (11.4% to 7.7% at low reasoning effort). API pricing is listed at $2 per million input tokens and $10 per million output tokens — matching GPT-6 Sol and Anthropic’s Claude Sonnet 5.5 on list rates, with cheaper cache reads, The Decoder (opens in new tab) reports. An “Ultrafast” Sol variant for Codex is expected within days.

On safety evaluations designed to be difficult, OpenAI says Sol tries to bypass explicit blocks less often than GPT-6 Sol (23.5% vs. 64.4%) but still more often than Astra (17.4%), and that neither Sol nor Astra attempted to circumvent an automated safety reviewer in the tests cited. Those figures are company-reported and not yet independently replicated.

The Sol release is OpenAI’s public product answer after shelving its louder flagship: keep a cheaper, somewhat safer workhorse in market while Astra’s base model undergoes more reinforcement-learning work. Competitive claims versus Claude Opus/Sonnet rest on OpenAI’s slides until third-party evals catch up.