Models Published:
NVIDIA's small Nemotron 3.5 Bolt E4B arrives in Ollama 0.40.10, with a DFlash draft for faster replies
What happened
Ollama v0.40.10, released on October 23, adds Nemotron 3.5 Bolt E4B, the smallest member of NVIDIA's Nemotron 3.5 family, with the command `ollama run nemotron-3.5-bolt:e4b`. NVIDIA post-trained it from Google's Gemma 4 E4B: it has 8B parameters, 4B of them effective, and is aimed at fast local conversations, instruction following and multi-turn tool use for voice and device assistants. Ollama lists image input and a 128K-token context window for its package, and the model comes with a DFlash draft model for speculative decoding. NVIDIA releases it under the OpenMDW 1.1 license and says it is ready for commercial use.
Why it matters locally: The Nemotron 3.5 family now has a member small enough for entry-level graphics cards and many laptops, not just 24 GB cards and workstations.
Three ways to read it
For the market
Small models around 4B effective parameters are a crowded field, and NVIDIA building on Gemma 4 E4B rather than training from scratch suggests it wants a quick foothold in local voice and device assistants. Being in Ollama the day it ships likely matters as much as the model itself.
In plain words
NVIDIA has released a small AI model meant for quick back-and-forth chats and simple tasks, and it is now a one-line download in Ollama, a free app for running AI on your own computer. It is small enough for many everyday computers, and a helper model can make its answers come faster.
For practitioners
Ollama lists the `e4b` tag at 7.2 to 7.9 GB, while NVIDIA's own 4 GB Q4_K_M GGUF card lists text-only input and up to 64K tokens, so check which build you are running before comparing results. Test the DFlash draft on your own prompts to see whether speculative decoding actually speeds up your hardware, and compare tool-call reliability with plain Gemma 4 E4B.
Sprout's insight
Sprout compares this with other brands and with where real applications are heading. Free for members.
Sign in free to read it →Subscribe to Sprout's weekly
One email a week: the top stories and a taste of Sprout’s insights. Unsubscribe any time.
Source
Read the official source: Ollama v0.40.10 release notes (GitHub) ↗Summaries are written by OnPrem World from the official source. We link to the original and never republish it.
On this site
More news
- Ollama 0.40.9 adds local text-to-speech on Apple Silicon, d3 decision models, and Flint on Linux and Windows
- Recursive releases Flint, an Apache 2.0 coding model with 40B parameters and about 2.5B active, already in Ollama for Mac
- AMD and Intel agree on oneDNN as PyTorch's shared x86 CPU acceleration library
- vLLM v0.32 brings 4-bit KV cache to more GPUs and stops allowing every CORS origin by default
- DeepSeek-V4.1-Flash-1017: a same-size update with better agent scores and shorter answers