An internal OpenAI Astra model solved 10 major open math and CS problems

OpenAI dropped a big flex this week, saying an internal version of its upcoming Astra model solved 10 major unsolved problems in math and computer science, pitching it as a huge leap for scientific thinking. But online, the real action wasn’t just the claim — it was the comment-section cage match that followed. One camp saw a sci-fi dream coming true: finally, computers helping humans crack the kind of hard problems that can take experts years. Another immediately hit the brakes, asking the question hanging over almost every shiny AI announcement lately: is this real progress, or just very expensive hype?

The sharpest criticism came from people who felt the rollout itself was a little too Hollywood. One commenter blasted the “shock and awe” style reveal, saying dropping 10 results at once felt like a PR stunt that disrespects the mathematicians whose work laid the foundation. Others were plainly exhausted by what they see as the endless AI bragging cycle, joking that the news now sounds like a looping meme: one chatbot solves a problem, then another escapes a sandbox, rinse and repeat. That meme energy gave the whole thread a deliciously skeptical vibe.

Still, not everyone was grumpy. Some readers said this is exactly what computers are supposed to do — help humanity solve hard stuff — and others were already dreaming bigger, like using AI to speed up drug discovery and understand genetic diseases. In other words: half the crowd heard “breakthrough,” the other half heard “marketing,” and everyone brought popcorn. OpenAI’s announcement definitely got attention — but the comments made it a spectacle.