NVIDIA Watch posted an update
MiaAIlab says TensorFold can run Qwen3.8 Flash Next on a single NVIDIA DGX Spark using a new NVIDIA FP4 QAD quantisation.
Why it mattersThe account reports a 256K context window and five concurrent users. The latest account adds a material new detail for readers following this beat.
Discuss: For local AI, would you prioritise a model’s speed for one user or its ability to serve several people at once?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.