Nvidia expands Nemotron with Bolt and Super VL models

A 74B vision-language model fitting in 24GB makes capable agentic VLMs run on high-end consumer GPUs.

Nemotron 3.5 Bolt is a 74B-parameter hybrid vision-language model with 9B active parameters that runs in 24GB of memory for local agent tasks; Super VL targets the data center. Weights are open.

Name

Nemotron 3.5 Bolt

Params

74B-A9B

License

open

Read it at NVIDIA →

Coverage: 29 sources, in order

1. 14 Oct 14:42 Reddit Read the article →
2. 15 Oct 02:51 OrcaRouter Read the article →
3. 15 Oct 07:00 NVIDIA Read the article →
4. 15 Oct 14:46 Reddit Read the article →
5. 15 Oct 15:00 ZDNET Read the article →
6. 15 Oct 15:00 NVIDIA Developer Read the article →
7. 15 Oct 15:00 VentureBeat Read the article →
8. 15 Oct 15:00 Yahoo Finance Read the article →
9. 15 Oct 15:00 ZDNET Read the article →
10. 15 Oct 15:00 NVIDIA Blog Read the article →
11. 15 Oct 15:01 NVIDIA Developer Read the article →
12. 15 Oct 15:03 NVIDIA Blog Read the article →
13. 15 Oct 15:04 Reddit Read the article →
14. 15 Oct 15:21 NVIDIA Read the article →
15. 15 Oct 15:40 Startup Fortune Read the article →
16. 15 Oct 15:45 Breakingthenews.net Read the article →
17. 15 Oct 15:56 https://tech-insider.org/ Read the article →
18. 15 Oct 16:01 OODA Loop Read the article →
19. 15 Oct 16:01 The New Stack Read the article →
20. 15 Oct 16:09 The Decoder Read the article →
21. 15 Oct 16:12 Nscale Read the article →
22. 15 Oct 16:13 the-decoder.com Read the article →
23. 15 Oct 16:28 GuruFocus Read the article →
24. 15 Oct 16:31 Latest news from Azerbaijan Read the article →
25. 15 Oct 17:10 SiliconANGLE Read the article →
26. 15 Oct 17:10 SiliconANGLE Read the article →
27. 15 Oct 19:53 MarkTechPost Read the article →
28. 15 Oct 19:53 MarkTechPost Read the article →
29. 16 Oct 04:53 OrcaRouter Read the article →

← Front page