Anyone else notice that the most capable models aren't actually available to us anymore?

via

OpenAI

There's a pattern that's been bugging me lately. The most powerful models being announced the ones with genuinely impressive benchmarks and specialized capabilities aren't being released to the public. They go into some kind of "trusted access" program or enterprise-only pipeline, and everyone else just reads the blog post. I get why labs are doing it. Safety concerns, dual-use risks, liability. It makes sense on paper. But it also means that the community doing the most interesting open-source

More from OpenAI

This is so cool. You can talk to an AI only trained on pre-1930 text. Really feels like talking to someone from the past.

OpenAI·15

why does GPT 5.5 have a restraining order against "Raccoons," "Goblins," and "Pigeons"?

OpenAI·14

"He wanted to be CEO": Early OpenAI VC Vinod Khosla says Elon Musk’s bid for control led to the Sam Altman feud and his major investment

OpenAI·13

Imagen 2.0

OpenAI·5

Sarah Friar strikes again!

OpenAI·12

OpenAI Models, Codex, and Managed Agents Come to AWS

OpenAI·7