NVIDIA Nemotron 3.5 Bolt E4B 4GB: an open small model for local voice agents, game characters and robots
NVIDIA released Nemotron 3.5 Bolt E4B 4GB on Oct 23, 2026: an open small language model post-trained from Google's Gemma 4 E4B with reinforcement learning only, packing conversational ability plus multi-turn tool use into about 4GB of VRAM for RTX, Jetson and Spark hardware.
Why it matters
It is built for the gap other small models leave: a chatty local agent that can also call tools across turns, on a single small device, with weights, data and recipes open under OpenMDW-1.1.
What you could build with it
A small robotics team could run Bolt E4B on a Jetson Orin Nano as the conversational brain of a tabletop assistant, pairing it with a vision model so the device talks, listens and triggers actions entirely offline.
Does it hold up?
Ready to run today via llama.cpp or Ollama on RTX/Jetson hardware, but NVIDIA's speed and benchmark claims are self-measured and the 4GB label refers to VRAM in target workloads, not the 6.18GB download.
Built with NVIDIA Nemotron 3.5 Bolt E4B 4GB
- NVIDIA Releases Nemotron 3.5 Bolt E4B 4GB: An Open Small Language Model for Local Voice Assistants, Game Characters and RobotsMarkTechPost · Launch breakdown: 8B total / 4B effective parameters, 64K context, and NVIDIA's benchmark claims against Gemma 4 E4B and Qwen3.5-4B.
- Nemotron 3.5 Bolt E4B Arrives, NVIDIA's Model for Robots and GamesPasquale Pillitteri · Launch report including the Reachy Mini demo — Bolt handling conversation on an 8GB Jetson Orin Nano while Cosmos handles vision.
- NVIDIA releases Nemotron 3.5 Bolt E4B for local voice agents and robotsDigest AI · Digest of the release noting the 4GB VRAM target, the 6.18GB GGUF download size, and distribution across Hugging Face, Ollama, LM Studio and Unsloth.
Learn more
First spotted on article: source.
More AI models
OpenAI publishes 722 AI-generated math manuscripts from an unreleased modelOpenAI released 722 mathematical manuscripts in 372 result families to GitHub under Apache-2.0, produced by an…models · JEV 0.74Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73NVIDIA Isaac Lab 3.0 GA + Isaac OS open runtime + Agile ONE SN humanoidMajor Isaac platform expansion at GTC Berlin Oct 21: Isaac Lab 3.0 is generally available as the open-source robot…models · JEV 0.73StepFun Step-5-Preview — open BF16 weights releasedStepFun published the full BF16 weights for its flagship Step-5-Preview agentic model (600B-parameter sparse MoE, 27B…models · JEV 0.73NVIDIA Nemotron 3.5 Retriever collection: Embed 8B/1B, Rerank 8B, Parse 631MNVIDIA's Oct 21 collection drop for agentic retrieval: Nemotron-3.5-Embed in 8B BF16, 1B BF16 and 1B NVFP4…models · JEV 0.72Kimi K3.1 Pro: Moonshot's 2.8T open-weight flagship officially launchesMoonshot AI officially launched Kimi K3.1 Pro on Oct 23, a mixture-of-experts model with 2.8 trillion total / 104…models · JEV 0.72
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.