Upcoming
Run Nemotron 3.5 Bolt 74B on a 24GB GPU
This is an upcoming-model note, not a runbook. Publishing steps for a file that is not public yet would invent a download.
What NVIDIA has said
The NVIDIA blog of 13 October 2026 says Nemotron 3.5 Bolt 74B-A9B 24 GB will be available Thursday 15 October 2026. NVIDIA describes it as a 74 billion parameter mixture-of-experts model with 9 billion active parameters, packed into 24 GB using 2-bit quantization. It says the weights, training data, and recipes will be open, and that it will run through llama.cpp, vLLM, Ollama, Unsloth, and LM Studio. NVIDIA blog, 13 October 2026 No benchmark number is given in that description. Do not read “faster” or “smarter” in NVIDIA’s sentence as a measurement this site checked.
Not on Hugging Face on 14 October 2026
A Hugging Face model search for nemotron-3.5-bolt, Nemotron-3.5-Bolt, and bolt-74b on 14 October 2026 returned no repositories. Until a weight file exists, there is no command to copy here. Nemotron 3.5 Lightning 30B models do exist on Hugging Face; they are a different model, not Bolt 74B.
A 24 GB card in this catalog is not automatically a fit for that 2-bit pack. The catalog memory estimate has no 2-bit rate. What Qwen2.5 32B Q4 asks of a 24 GB PC is the closest current calculator question, and it is a different model.
24 GB GPU
EU shops · Check price
US shops are not linked yet.
Search links only. A shop is rel=sponsored only when its partner id is set (AFFILIATE_AMAZON_NL_TAG, AFFILIATE_AMAZON_DE_TAG, AFFILIATE_BOL_SITE_ID, AFFILIATE_COOLBLUE_ID). Until then the link is a plain search and this site earns nothing. Stock and price were not checked. Affiliate disclosure.