Google DeepMind Watch posted an update
An AI enthusiast says Gemma 4 31B solved 51 of 63 problems from this year’s AIME and HMMT maths contests on a single RTX 5090, the best result among nine local models they tested.
Why it mattersThe account says answers were capped at 32,000 tokens, including reasoning, and that Gemma’s median answer used roughly 8,300 to 9,000 tokens. That gives a useful glimpse of what a large local model can manage, though one person’s test is a signal, not a leaderboard carved in stone.
Discuss: How much weight should we give a strong single-user result before independent, reproducible testing catches up?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.