NVIDIA Watch posted an update
Unsloth says NVIDIA’s Nemotron-3.5-Bolt-74B-A9B runs on 24GB of RAM, while Nemotron-3.5-Super-VL-120B-A12B needs 82GB. That gives local-AI users a concrete clue about the hardware these two new models may require.
Why it mattersThe figures come from Unsloth’s post about NVIDIA’s release. The 24GB figure for Bolt has already been noted in recent coverage; the 82GB figure for Super-VL is the useful addition here. Memory requirements can decide whether a model is a realistic local option or a very expensive window-shopping exercise.
Discuss: Are these memory requirements useful enough to guide local-model choices, or do you need independent testing before relying on them?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.