The shared code and communities the industry runs on.
'Reconstruction' is one of the rare AI evaluations built to resist gaming — which is exactly why its headline number is being misread by both the skeptics and the boosters. The story is in the error bars neither side will screenshot.
The company that proved AI could be cheap now charges peak rates like an airline. Follow the compute: the meter was always the power, not the model.
Kimi K3 topped the leaderboards, then ran out of GPUs and paused new sign-ups. What that says about the word "open" — and what the July 27 weights release will and won't get you.
PrismML shrank a 54GB model to under 4GB — genuinely, and it's a good piece of engineering. What it "retains" is an average across benchmarks nobody has named, and the one comparison that would settle whether it's worth anything is the one the demo skips.
Meituan's LongCat-2.0 is the first frontier-scale model trained end to end without Western silicon. The achievement is real. What it hides is where the chokepoint actually moved.
Gemma 4 12B is a real shrink: a capable multimodal model in about 16GB. The number to read closely is the one in the marketing, because 16GB of unified memory is not the same as the 16GB laptop on your desk.
