Skip to content
Log in

nvidia/Cosmos-Reason2-8B

huggingface_fp16_tensorrt-edgellm
nvidia/Cosmos-Reason2-8B architecture graph
nvidia/Cosmos-Reason2-8B architecture graph
  • huggingface
  • AGX Thor
Labels
  • Image
  • Text
  • VLM
  • Generation
Build
Embedl-optimized
Precision
W4A16
Runtime
TensorRT Edge-LLM 0.10.1 with Embedl optimizations
Context length
4,096 tokens

NVIDIA's 8B Cosmos-Reason2 vision-language model for physical-world reasoning in robotics and video analytics.

Files

Image Size

Model

docker.io/embedl/model-artifacts:cosmos-reason2-8b.huggingface_int4_embedl_tensorrt-edgellm.16e317c04d4a006c

 6.78 GiB

Runtime

docker.io/embedl/models:runtimes.tensorrt-edgellm.arm64.2026101815014ee0

 1.01 GiB