OpenAI's New GPT-6.1 Sol Brings Flagship AI at One-Fifth the Price
OpenAI has released GPT-6.1 Sol, an upgraded version of its GPT-6 Sol model that the company says delivers intelligence close to its most powerful system, GPT-6 Astra, at roughly one-fifth of the price. The launch, reported by Chinese financial media including Cailianshe, marks another step in the industry’s push to make advanced artificial intelligence cheaper and more widely usable.
The pricing is the headline. GPT-6.1 Sol is listed at two dollars per million input tokens and ten dollars per million output tokens, with cached input tokens at ten cents per million. By comparison, GPT-6 Astra is priced at ten dollars for input and fifty dollars for output per million tokens. For companies building products on top of large language models, that gap can decide whether an application is commercially viable or not.
OpenAI says the new model performs especially well in agentic coding, computer use and professional work — tasks where the AI is asked to carry out multi-step jobs rather than simply answer questions. That focus reflects a broader shift in the industry: the most valuable AI systems are increasingly judged by what they can do autonomously, not just how fluently they chat. The company also said an ultra-fast version of GPT-6.1 Sol will arrive in the coming days.
Alongside the model, OpenAI introduced Dots, described as a round-the-clock autonomous agent. The idea is an AI that keeps working in the background — handling tasks, monitoring information and taking actions — rather than waiting for a user to type a prompt. Autonomous agents have become one of the most competitive battlegrounds in AI, with major labs racing to build systems that can reliably complete real work without constant supervision.
The timing is notable. According to people cited in the same reports, OpenAI’s annualized recurring revenue is now approaching 70 billion dollars, up more than 70 percent since early in the third quarter, with enterprise sales more than doubling since July. Those figures, if accurate, would underline how quickly businesses have moved from experimenting with AI to paying for it at scale. The reports also noted that the company’s spending picture remains less clear.
The announcement lands in a global AI market that is both booming and nervous. Demand for computing power has driven up prices and investment in memory chips, data centers and specialized hardware, while investors increasingly debate how long the spending surge can last. A cheaper, nearly flagship-level model from a leading lab could accelerate adoption — and intensify pressure on rivals to match both performance and price.
For ordinary users, the practical effect may be subtle but real: more capable AI assistants embedded in everyday software, lower costs for developers, and a growing number of services that quietly run in the background. For businesses, the calculus is simpler. When frontier-level intelligence costs a fifth as much, the list of tasks worth automating gets longer.
OpenAI has not said exactly when the ultra-fast variant will ship or how Dots will be priced. But the direction is clear. The race in AI is no longer only about who builds the smartest model — it is about who can deliver that intelligence cheaply enough, and reliably enough, to become part of daily work.