Cosmos-Predict2.5-14B
No estimate
No hardware requirements for this model
The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.
On record
Full specification
Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.
Origin
Who built this model, where, and when it was published.
- Organisation
- NVIDIA
- Organisation type
- Industry
- Country
- United States of America
- Published
- 29 September 2025
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Video
- Task
- Video generation, Text-to-video, Image-to-video, Video-to-video
Size
How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.
- Parameters
- 14B
- Training data
- tokens
"Trained on 200M curated video clips and refined with reinforcement learning–based post-training" "we apply the same 1 × 2 × 2 patchification strategy to compress latent features further. We train our model to generate 93 frames, which corresponds to 24 latent frames, at a time using 16 fps videos. Each of the generated videos is about 5.8 seconds long." 200*10^6 * 5.8 / 3600 = 322222 hours of video
The training run
What it physically took to train: which chips, how many, for how long, and what that drew from the wall.
- Training hardware
- NVIDIA H100 SXM5 80GB
- Chips used
- 4,096
- Hardware utilisation
- MFU 33.1%
- Power draw
- 5.6 MW
Availability
Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.
- Weights
- Closed — provider access only
- Model access
- Unreleased
- Training code
- Unreleased
How it is classified
Labels the source dataset applies when tracking notable models, and how confident it is in the entry.
- Record confidence
- Confident
Sources
Where this record came from and when it was last checked.
- Reference
- World Simulation with Video Foundation Models for Physical AI
- Last updated
- 28 November 2025
What the numbers mean
Where it came from
Cosmos-Predict2.5-14B was published by NVIDIA, in United States of America, in September 2025. industry is the category the publisher falls under.
It works in Video, and is recorded as doing video generation, Text-to-video, Image-to-video, Video-to-video.
This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.
Answers
Cosmos-Predict2.5-14B — common questions
What GPU do I need to run Cosmos-Predict2.5-14B?
None. Cosmos-Predict2.5-14B is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.
Is Cosmos-Predict2.5-14B open source?
No. Cosmos-Predict2.5-14B has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does Cosmos-Predict2.5-14B have?
Cosmos-Predict2.5-14B has 14B parameters. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.
Who created Cosmos-Predict2.5-14B?
Cosmos-Predict2.5-14B was published by NVIDIA, based in United States of America, categorised as industry.
When was Cosmos-Predict2.5-14B released?
Cosmos-Predict2.5-14B was published in September 2025.
What is Cosmos-Predict2.5-14B used for?
Cosmos-Predict2.5-14B works in Video, and is recorded as handling video generation, Text-to-video, Image-to-video, Video-to-video. These are the areas it was designed around; they describe intent rather than a hard boundary.
The other direction
Looking at it from the other side?
This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.