Nvidia ships open agent models for its own RTX Spark laptops
The chipmaker now sells the PC platform, the chip and the open model that runs on it. Microsoft's Surface Laptop Ultra leads the first wave.
Di The desk, News · · 1 min di lettura
Tag: nvidia · open weights · laptops · agents
Questo pezzo non è ancora in Italiano. Stai leggendo l'inglese.
- La notizia
- The chipmaker now sells the PC platform, the chip and the open model that runs on it.
- Perché conta
- Microsoft's Surface Laptop Ultra leads the first wave.
- Cosa è confermato
- The photograph is generated for this story.
Legato a una fonte citata. Non segnato come verificato. NVIDIA Newsroom
The news
Nvidia released Nemotron 3.5 Super VL, a mixture-of-experts vision-language model with 120 billion total and 12 billion active parameters, with open weights, datasets and training recipes. Alongside it came two smaller Bolt models aimed at RTX Spark laptops from Microsoft, Asus, Dell, HP, Lenovo and MSI.
Why it matters
Nvidia is pairing open models with its own Arm-based Windows platform in launch week. Local AI agents on the desk are now a race between Apple's Macs and Nvidia's PCs.
The details
Nvidia says Super VL matches its largest reasoning model on agent tasks at about a fifth of the size. Those are company claims. The independent Artificial Analysis index scores the model 26, above the median for open models of similar size.
The Bolt 74B-A9B model is quantised to fit in 24 GB of memory and is available now; a 4 GB version is due on October 23. The Surface Laptop Ultra starts at 2,599 dollars with up to 128 GB of unified memory, Tom's Hardware reported. vLLM added same-day support for Super VL.
What we could not check
- We have not run the models ourselves.
- Speed comparisons cite Nvidia's selection of third-party data.
Fonti
The photograph is generated for this story. It does not show the people named in it.