MODELS

GPT-6 Sol

The cost-efficient high-end tier of OpenAI’s GPT-6 series, between Astra and Luna, built for complex coding and agentic workflows.

Released
22 Sep 2026
Developer
OpenAI
Context window
1.05M tokens
Modalities
  • Text
  • Image
Pricing / 1M
$2 in · $10 out
Reasoning depth
  • Low
  • Medium (used for these runs)
  • High
  • xHigh
PROMPT
353 WORDS

Mass Intelligence Needs Mass Agency

A billion people with powerful AI will not automatically make a wiser world. Intelligence is the ability to solve problems; wisdom is choosing which problems deserve solving, and who gets to live with the consequences. My strongest conviction is that the central danger of Mass Intelligence is not that machines will think too much. It is that a few institutions will decide too much while everyone else gets a very clever interface.

We should treat AI as public infrastructure, not merely as a product. That means affordable access, independent research, open standards, and the right to move our data and work between providers. It does not mean publishing every dangerous capability. A system that can design a better lesson plan should be widely available; one that can help engineer a pathogen needs rigorous controls. Openness is a tool, not a religion.

We also need to change what we reward. If AI makes persuasion nearly free, attention becomes a battlefield. Platforms should be liable for predictable harms from mass-produced deception, political ads should disclose synthetic content, and people should have the practical ability to verify where consequential claims came from. Schools should teach students to argue with AI, test its answers, and recognize when it is flattering them. The goal is not to produce better prompt writers. It is to produce harder-to-fool citizens.

At work, productivity gains should buy people bargaining power and time, not just cheaper labor for employers. Workers need a say before AI systems measure, schedule, or replace them. Governments should fund transitions and capture some of the windfall through ordinary, enforceable taxation. Otherwise we will call it progress while making millions feel disposable.

Finally, keep humans accountable. No model should be the last word on a prison sentence, a medical denial, or a military strike. Require an identifiable person with the authority, information, and obligation to say no.

Mass Intelligence will help us thrive only if we build institutions that spread its benefits, contain its hazards, and protect our capacity to decide together. The scarce resource is no longer answers. It is democratic power over what we do with them.

GPT-6 Sol | productized.tech