Skip to content
RobothecaBeta

Cosmos-Transfer2.5

  • AI model
  • Humanoids and robot learning

A multi-controlnet video generation model from NVIDIA that turns structured video inputs such as RGB, depth and segmentation into new generated video for physical AI and robotics data augmentation.

  • Custom weights terms

Made by NVIDIA, United States (checked on github.com/nvidia-cosmos (opens in a new tab)).

Source on GitHub: github.com/nvidia-cosmos/cosmos-transfer2.5 (opens in a new tab). Facts read at commit 2ff49d0 (opens in a new tab).

Licenses

Code
Apache-2.0 (opens in a new tab)

Read by Robotheca from LICENSE (opens in a new tab) (Apache-2.0) at commit 2ff49d0 (opens in a new tab), on 5 Nov 2026.

Weights
NVIDIA Open Model LicenseTerms of their own, not a standard license

Read on huggingface.co/nvidia/Cosmos-Transfer2.5-2B (opens in a new tab).

Commercial use
Unknown

A license Robotheca cannot classify by itself: read it before any commercial use.

Labels

  • Custom weights termsThe weights come under their maker's own terms (NVIDIA Open Model License), not a standard license: read their conditions before any use.

What it does: Robotheca's summary

  • Accepts several video modalities as control inputs and is configured through JSON controlnet_specs.
  • Supports single-video inference, automatic control map generation, and multiple GPU setups.
  • Offers Simulation-to-Real and Real-to-Real augmentation examples, including weather and scene changes.
  • Includes checkpoints for general robotics use and for autonomous vehicle and robot multiview applications.
  • Provides distilled checkpoints for low-latency edge deployment, plus image-to-image and image prompt options.

Written by Robotheca from the project's README (opens in a new tab) at commit 2ff49d0.

Getting started

Difficulty
Advanced

Maintenance

Status
Maintained: a commit in the last 12 months
Newest commit
7 Jun 2026 on main
Latest release
v1.5.4 (13 May 2026)

Checked on GitHub on 5 Nov 2026.

Links

Robots that can run it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.