ModelsPublished All news from that day
Nvidia released Nemotron 3.5 Bolt, a 74B open-weight vision-language model that fits in 24GB, under the OpenMDW licence.
Nvidia's own benchmark table lists 50.98 on Terminal-Bench 2.1 and 66.75 on BrowseComp.
Check our sources · 7 facts from 1 sourceNvidia's model card for NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB gives a release date of 10/15/2026 on Hugging Face. The card describes a vision-language model with 74B total and 9B active parameters, designed to run everyday agentic tasks locally. It uses a native 2-bit quantization to fit in 24GB of memory.
Context length is up to 256K tokens. Nvidia says it compressed the model from the Nemotron 3.5 Super VL parent checkpoint of 120B-A12B. Use is governed by the OpenMDW License Agreement, version 1.1. On Nvidia's own benchmark table the model scores 50.98 on Terminal-Bench 2.1 and 66.75 on BrowseComp.
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against this source.
1 NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB model card
Open the source archived copy-
The model card for nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB gives its release date as Hugging Face 10/15/2026.
citeHugging Face 10/15/2026
-
The model card says Nemotron 3.5 Bolt 74B-A9B 24GB is a vision-language model designed to run everyday agentic tasks locally, using a native 2-bit quantization to fit in 24GB of memory.
citea native 2-bit quantization to fit in 24GB of memory
-
The model card lists 74B total parameters (9B active).
cite74B total parameters, with 9B active
-
The model card says the model supports up to 256K tokens of context.
citeup to 256K tokens
-
The model card says Nvidia compressed the Nemotron 3.5 Super VL parent checkpoint (120B-A12B) to Nemotron 3.5 Bolt 74B-A9B 24GB using Star Elastic.
citeWe compress the Nemotron 3.5 Super VL parent checkpoint (120B-A12B) to Nemotron 3.5 Bolt 74B-A9B 24GB using Star Elastic
-
The model card says use of the models is governed by the OpenMDW License Agreement, version 1.1.
citeOpenMDW License Agreement, version 1.1
-
The model card's benchmark table, reported by Nvidia, lists Nemotron 3.5 Bolt (IQ2) at 50.98 on Terminal-Bench 2.1 and 66.75 on BrowseComp. the company says
citeTerminal-Bench 2.1: Bolt (IQ2) 50.98; BrowseComp: Bolt (IQ2) 66.75
Every sentence above is checked against this source. Each fact has its own address you can cite.
Source 1 · NVIDIA (Hugging Face) · 15 Oct 2026
NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB model card
The model card for nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB gives its release date as Hugging Face 10/15/2026.
Hugging Face 10/15/2026
The model card says Nemotron 3.5 Bolt 74B-A9B 24GB is a vision-language model designed to run everyday agentic tasks locally, using a native 2-bit quantization to fit in 24GB of memory.
a native 2-bit quantization to fit in 24GB of memory
Show all 7 factsShow fewer facts
The model card says Nvidia compressed the Nemotron 3.5 Super VL parent checkpoint (120B-A12B) to Nemotron 3.5 Bolt 74B-A9B 24GB using Star Elastic.
We compress the Nemotron 3.5 Super VL parent checkpoint (120B-A12B) to Nemotron 3.5 Bolt 74B-A9B 24GB using Star Elastic
The model card says use of the models is governed by the OpenMDW License Agreement, version 1.1.
OpenMDW License Agreement, version 1.1
The model card's benchmark table, reported by Nvidia, lists Nemotron 3.5 Bolt (IQ2) at 50.98 on Terminal-Bench 2.1 and 66.75 on BrowseComp. the company says
Terminal-Bench 2.1: Bolt (IQ2) 50.98; BrowseComp: Bolt (IQ2) 66.75
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep