NVIDIA releases 2-billion-parameter Alpamayo 2 Edge weights on Hugging Face for autonomous driving
NVIDIA released the Alpamayo 2 Edge model weights on Hugging Face on November 10, 2026. The 2-billion-parameter model targets autonomous-driving developers working within onboard compute and latency limits, and ships with deployment recipes and evaluation tools.
NVIDIA released the Alpamayo 2 Edge model weights on Hugging Face on November 10, 2026, according to Unite.AI. The model has 2 billion parameters and is aimed at autonomous-driving developers who work within onboard compute and latency limits.
What is included
The release includes deployment recipes and evaluation tools. According to the report, these are meant for producing and scrutinizing driving trajectories and their explanations.
Background
NVIDIA first announced Alpamayo 2 Edge in its October 21, 2026, Hyperion announcement. That announcement described a model optimized for DRIVE AGX Thor.
Specifications
- Parameters: 2 billion
- Distribution: model weights on Hugging Face
- Release date: November 10, 2026
- Context window, pricing and benchmark scores: not yet disclosed
What this means
Publishing the weights, rather than only announcing the model, lets developers obtain Alpamayo 2 Edge directly from Hugging Face. The accompanying deployment recipes and evaluation tools show that NVIDIA is positioning the release for work on driving trajectories and their explanations under onboard compute and latency constraints. The 2-billion-parameter size is the figure NVIDIA has attached to that edge target.
Related Articles
NVIDIA Alpamayo 2 Edge: 2B-Parameter Driving Model Hits 160-190 ms Latency on DRIVE AGX Thor
NVIDIA has introduced Alpamayo 2 Edge, a 2B-parameter adaptive reasoning vision-language-action model built for real-time operation on DRIVE AGX Thor. NVIDIA says deployment recipes with TensorRT and DFlash2 reach 160-190 ms pixel-to-trajectory latency, and that closed-loop RL improved driving performance by 48.9%.
MiniMax releases H3-Image: 33.1B-parameter model generates 16-megapixel images in 6 sampling steps
MiniMax has released MiniMax-H3-Image, a 33.1B-parameter diffusion transformer that handles text-to-image and image editing in one checkpoint. It generates images of about 16 megapixels directly, without an upscaler, using 6 sampling steps and no classifier-free guidance.
Underdog Woof, a 40.6 MB two-voice TTS model, claims 1.42% word error rate on a CPU
Underdog Woof is an English text-to-speech model with two voices, Juno and Mira, packaged as a single 40.6 MB file that runs on a CPU with numpy. Underdog reports a 1.42% word error rate for Mira on Gradium TTS Hard Cases, against 1.89% for Fish Audio Drama 3, a 41.5 GB download.
Video Rebirth releases HyperFlow H3 Flash, a 14B few-step video and audio model compressed from MiniMax-H3
Video Rebirth has published HyperFlow H3 Flash, a compressed, few-step version of MiniMax-H3 with 13.7B to 13.9B parameters. It generates 5.2 to 15 second videos at 24 fps with synchronized 32 kHz stereo audio, using 4, 6 or 8 sampling steps without classifier-free guidance.
Comments
Loading...