Testing NVIDIA Cosmos: Accurate Trajectories but Fails Object Counting
MaziyarPanahi · x · 2026-08-09
A developer created a mini-game using NVIDIA's open-source Physical AI model, Cosmos, to test its capabilities. The setup involves asking the model to predict the next 3 seconds of footage, rolling the actual video, and letting the model grade its own performance.
The test revealed that while Cosmos accurately predicted the motion of jets, it failed at basic object counting—predicting six jets when there were only four. The demo was run locally on a Mac Studio using MLX and SAM 3.1.
Related event: Testing NVIDIA Cosmos: Accurate Trajectories but Fails at Counting(4 posts)→
More from Models
- GYRO quant of Qwen3.8-Flash-Next tested: near Q3 quality at much smaller size — PlasticRevenue4601 · 2026-10-16
- Open source is so back: dozens of open models dropped within 48 hours — beffjezos · 2026-10-16
- Gemini users hit usage caps on 3.1 Pro and 4 Argon, resets not until Oct 16 — R3d-Gr33n-Blu3 · 2026-10-16
- Gemini desktop app can't accept images — an image model that can't take images — Rebl · 2026-10-16
- Apollo Research: GLM-5.3 unprompted reasons about its actions becoming training data — _lewtun · 2026-10-16
- Opus 5.5 turns heads with surprisingly polished vertical video demo — FinanceYF5 · 2026-10-16