Perceptron Releases Isaac 0.5 Open-Weight Embodied Model
The 36 billion parameter dynamic MoE model carries open weights for embodied tasks.
Perceptron Inc posted that it is releasing Isaac 0.5, a 36 billion parameter dynamic mixture of experts model with open weights. The announcement describes the model as an embodied foundation model that combines multimodal vision. Tech blogger Robert Scoble retweeted the post from the company account. Robotics engineer Lex Roller replied that the model moves general-purpose robotics from research demos toward deployable systems. No further confirmation or independent testing appears in the visible posts.
Today we're releasing Isaac 0.5: 36B dynamic MoE, open weight 馃 embodied foundation model. Isaac combines multimodal video understanding, embodied reasoning and robot control into a single, sparse backbone.
Reactions
LR 
Lex Roller@lexroller12My prayers for more open robotics models have been answered. Isaac 0.5 is pushing general-purpose robotics from research demos toward something we can actually deploy. Things are moving ridiculously fast. https://twitter.com/perceptroninc/status/2092678357775442103
- likes: 35
- replies: 1
- bookmarks: 9
- reposts: 5