Model creator

NVIDIA

A US technology company whose research teams publish the open-weight Nemotron model family. Its separate hardware and inference-platform roles are not recorded here as creator facts.

3 familiesChecked Aug 28, 2026

Featured ecosystems

Recorded lineage, creator by creator

Nesting shows a recorded predecessor Model Passport

CreatorNVIDIAA US technology company whose research teams publish the open-weight Nemotron model family. Its separate hardware and inference-platform roles are not recorded here as creator facts.

Family

Nemotron Nano 2

First released Aug 18, 2025

No recorded lineage links in this family, so no releases are connected here.

No creator statement of what this variant name means is recorded for Nemotron Nano 2.

Family

Cosmos

First released Jan 6, 2025

No recorded lineage links in this family, so no releases are connected here.

No creator statement of what this variant name means is recorded for Cosmos.

Family

Nemotron-4 340B

First released Jun 14, 2024

No recorded lineage links in this family, so no releases are connected here.

No creator statement of what this variant name means is recorded for Nemotron-4 340B.

Families without a featured release are not drawn here. Every record in the catalog, featured or not, is listed in the model index and the Model Tree.

NVIDIA / Nemotron Nano 2 Verified

Nemotron Nano 9B v2

A compact reasoning model that NVIDIA released on 18 August 2025 and documents as taking text input and producing text output, with a stated context length of up to 128K. Use is governed by the NVIDIA Open Model License Agreement, which does not appear on the Open Source Initiative's published index of approved licences. Weights are recorded as downloadable on the Hugging Face Hub API record, which reports the repository as ungated and enabled and lists its safetensors shards; the licence name alone was not treated as establishing that. Status is unknown because the card states no lifecycle term. The context window records 128000, the decimal reading of the card's bare "128K", the only figure it gives. Parameter count is not recorded: the card states none in prose and the "9B" appears only in the model name.

Release: Aug 18, 2025
Status: Unknown
Access: Open-weight

Recorded lineage

Earlier in lineage: No recorded predecessor.
Later in lineage: No recorded successor.
Released alongside: No recorded sibling release.
Derived from: No recorded derivation.

Lineage trail

Focus dims unrelated releases and lists every recorded relationship.

When to use it

NVIDIA's card describes the model as suited to developers designing AI agent systems, chatbots and retrieval-augmented generation systems, and to typical instruction-following tasks, across the ten languages the card names for input. NVIDIA states the model is distributed for download under its Open Model License Agreement.

View Model Passport Add to comparison Primary source

NVIDIA-Nemotron-Nano-9B-v2 model card | Checked Sep 6, 2026

Official record

The organization's own surfaces, and the verification date these facts were last checked against them.

Website: https://www.nvidia.com/
Release page: https://blogs.nvidia.com/
Type: company
Verified: Aug 28, 2026

Releases

Every release recorded for NVIDIA, current and legacy alike. Filter by lifecycle status; the choice is kept in the page address so it survives a reload or a shared link.

Showing all 3 releases.

Nemotron Nano 9B v2 Unknown
Nemotron Nano 29B Aug 18, 2025 Open-weight
Cosmos 1.0 Diffusion 7B (Text2World) Unknown
CosmosDiffusion 7B Text2World Jan 6, 2025 Open-weight
Nemotron-4 340B Base Available
Nemotron-4 340BBase Jun 14, 2024 Open-weight

Primary sources

The organization-level sources behind the summary above.

official announcementNVIDIA Releases Open Synthetic Data Generation Pipeline with Nemotron-4 340B
NVIDIA | Checked Sep 11, 2026
model cardNemotron-4-340B-Base model card
NVIDIA | Checked Sep 11, 2026