Models Published:
Ollama 0.40.5 adds NVIDIA Nemotron 3.5 Super and the 24 GB Nemotron 3.5 Bolt
What happened
Ollama 0.40.5 adds NVIDIA Nemotron 3.5 Super to its library, a vision-language model for coding, reasoning, tool use and document understanding. It has 120B parameters with 12B active per token in a hybrid Mamba-2/Transformer mixture-of-experts design, plus tool calling, an on/off thinking mode and context up to 1M tokens. A smaller sibling, Nemotron 3.5 Bolt, has 74B parameters with 9B active, is tuned to fit a 24 GB NVIDIA GPU and includes a drafter for speculative decoding.
Why it matters locally: Both models are now one `ollama run` command away, and Bolt is aimed squarely at the 24 GB cards many people already own.
Three ways to read it
For the market
NVIDIA is pushing its own open models into the most popular local runtime, which likely helps it compete with Qwen and Gemma for the default model on a home or office GPU. A 74B model sized for 24 GB could also make existing 24 GB cards feel more capable instead of pushing buyers toward bigger ones.
In plain words
Ollama, a free app for running AI on your own computer, can now download two new NVIDIA models that can read images and documents as well as text. The smaller one, Bolt, is made to fit on a single high-end graphics card with 24 GB of memory.
For practitioners
Bolt is tuned to fit a 24 GB NVIDIA card, so test it on your own tasks against the models you already run and check how much context still fits next to the weights. For the 120B Super model, compare the download size in the Ollama library with your GPU or unified memory before pulling it; this release also updates the bundled llama.cpp and MLX.
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 v0.40.5 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
- llama.cpp v0.7.0 adds a GPU cache for offloaded MoE experts and moves the server to port 9931
- NVIDIA RTX Spark laptops with up to 128 GB of unified memory go on sale October 16
- AMD's next Ryzen AI Halo: Ryzen AI Max+ PRO 495 with up to 192 GB of unified memory
- Open WebUI 0.12 adds two-factor sign-in, shared chats you can reply in, and realtime voice
- Ollama 0.40.2 upgrades older model downloads in the background and keeps backups