In recent weeks, two major AI companies have introduced models that are much cheaper to use than their flagship models.
GPT-6.1 Sol from OpenAI
OpenAI released GPT-6.1 Sol on 29 September 2026. According to media reports:
- It costs $2 per million input tokens and $10 per million output tokens, about a fifth of the price of the company’s stronger model, GPT-6 Astra.
- OpenAI says it comes close to Astra in programming, computer use and professional work.
- It is available in ChatGPT on paid business plans and in the API for developers.
Gemini 4 Argon from Google
On 13 October 2026, Google published the Gemini 4 Argon page in its developer documentation. According to media reports, its introductory price matches GPT-6.1 Sol and will double from late November. Reports differ on when the model will be generally available.
For small businesses this makes customer replies, document summaries and agents cheaper to run. The price per token is not the whole story, though: the real cost depends on how many tokens each task uses, so it is worth testing your own task on a few models.
To compare these tools, see Claude, ChatGPT, Gemini, Grok and Codex compared, and for building an agent for your business, AI agent development.
Photo: BalticServers.com, licensed under CC BY-SA 3.0, via Wikimedia Commons; cropped.
Source: TechCrunch ↗
