NVIDIA ships 4GB Nemotron 3.5 Bolt that runs locally on Jetson Orin Nano 8GB, eyed for Reachy mini robots
alexcovo_eth · x · 2026-10-24
NVIDIA released Nemotron 3.5 Bolt E4B 4GB on the redesigned Jetson AI Lab 2.0: a compact conversational model (8B total / 4B effective parameters) aimed at local assistants, multi-turn tool use, and interactive characters. It has a 64K context, OpenMDW-1.1 license, and ships as Q4KM GGUF, runnable via llama.cpp across the Jetson lineup from Orin Nano 8GB up to AGX Thor 128GB, with OpenAI-compatible serve/client commands provided. The poster, awaiting a Reachy mini robot, plans to buy a Jetson Orin Nano 8GB to run it — a low-cost combo of small robots with on-device models.
Related event: NVIDIA open-sources 4GB edge model Nemotron 3.5 Bolt E4B(11 posts)→
More from Embodied
- OpenTac: JHU/MIT open-source tactile sensing ecosystem with shear force and auto CAD — _krishna_murthy · 2026-10-25
- Rabbit exec: we're the only real AI-native hardware player, 100k+ units sold — SimonBalmain · 2026-10-25
- AGI House hosts Google Gemma hackathon with Arduino Ventuno Q hardware — agihouse_org · 2026-10-25
- print3d: open-source agent skill turns photos or text into Bambu Lab 3D prints — udmrzn · 2026-10-25
- Thore Graepel launches Metis Reasoning for embodied AI reasoning — SamuelAlbanie · 2026-10-25
- SimReady asset ecosystem for robot simulation keeps growing — jonstephens85 · 2026-10-25