Cosmos-Predict2 2B Video2World on Apple M2 Pro (32 GB)

Can the Apple M2 Pro (32 GB) run Cosmos-Predict2 2B Video2World?

53 pts

Yes, with little memory to spare

Tight fit. At FP8 it needs about 20.5 GB of 21.4 GB usable, with an estimated speed of 60.3 min.

Best precision: FP8
Memory mode: Sequential CPU offload
Time per clip: 60.3 min
Output: 1280×704, 93 frames, 35 steps

Download Cosmos-Predict2 2B Video2World

Original weights · Hugging Face

Full-precision files from NVIDIA, for fine-tuning, vLLM, Transformers, Diffusers or ComfyUI.

huggingface.co/nvidia/Cosmos-Predict2-2B-Video2World (opens in a new tab)

Free Hugging Face account needed: accept the license on the model page and the download starts right away.

Official repo · GitHub

Reference code, inference scripts and docs from NVIDIA. The weights themselves download from Hugging Face.

github.com/nvidia-cosmos/cosmos-predict2 (opens in a new tab)

git clone https://github.com/nvidia-cosmos/cosmos-predict2

Links checked 6 days ago against Hugging Face, Ollama and GitHub.

How the C grade is scored

Every result gets up to 100 points from four parts. The points decide the letter grade.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	25.6 / 45
Memory fitFull marks when everything sits in GPU or unified memory; fewer when part spills to system RAM.	10.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	3.3 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	15.0 / 15
Total	53.9 / 100	53 ptsUsable (C) · 53 of 100 points: Model quality 25.6/45, Memory fit 10.0/25, Speed 3.3/15, Memory use 15.0/15

Grade scale: S 85+ · A 72+ · B 58+ · C 42+ · D 25+ · F under 25 points.

Every precision on the Apple M2 Pro (32 GB)

Precision	Memory needed	Available	Time per output	Memory mode	Fit	Grade · points	Download
`FP16`	21.2 GB	21.4 GB	—	—	Won't run	7 ptsWon't run (F) · 7 of 100 points: Model quality 7.8/45, Memory fit 0.0/25, Speed 0.0/15, Memory use 0.0/15	Original weights (opens in a new tab)
`FP8`	20.5 GB	21.4 GB	60.3 min	Sequential CPU offload	Tight fit	53 ptsUsable (C) · 53 of 100 points: Model quality 25.6/45, Memory fit 10.0/25, Speed 3.3/15, Memory use 15.0/15	Original weights (opens in a new tab)

Other hardware that runs it

80 ptsGreat (A) · 80 of 100 points: Model quality 25.6/45, Memory fit 25.0/25, Speed 14.7/15, Memory use 15.0/15 NVIDIA GeForce RTX 5090 FP8 · 3.1 min
74 ptsGreat (A) · 74 of 100 points: Model quality 26.1/45, Memory fit 25.0/25, Speed 14.9/15, Memory use 8.1/15 NVIDIA RTX PRO 6000 Blackwell FP16 · 3.1 min
71 ptsGood (B) · 71 of 100 points: Model quality 25.6/45, Memory fit 18.0/25, Speed 13.1/15, Memory use 15.0/15 NVIDIA GeForce RTX 4090 FP8 · 3.9 min
71 ptsGood (B) · 71 of 100 points: Model quality 26.1/45, Memory fit 25.0/25, Speed 5.3/15, Memory use 15.0/15 Apple M4 Max (64 GB) FP16 · 24.4 min
70 ptsGood (B) · 70 of 100 points: Model quality 25.6/45, Memory fit 18.0/25, Speed 12.4/15, Memory use 15.0/15 AMD Radeon RX 7900 XTX FP8 · 4.4 min
70 ptsGood (B) · 70 of 100 points: Model quality 26.1/45, Memory fit 25.0/25, Speed 4.2/15, Memory use 15.0/15 Apple M1 Max (64 GB) FP16 · 37.5 min

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.