---
license: other
license_name: nvidia-nemotron-open-model-license
license_link: https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-nemotron-open-model-license/
base_model: nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4
base_model_relation: quantized
pipeline_tag: text-generation
library_name: gguf
tags:
  - nvidia
  - nemotron
  - nemotron-3
  - nemotron-3-super
  - gguf
  - q4_k_m
  - yarn
  - long-context
  - llama.cpp
  - ollama
language:
  - en
  - fr
  - es
  - it
  - de
  - ja
  - zh
---

# Nemotron-3-Super — Q4_K_M GGUF with YaRN 1M (Ollama / llama.cpp)

Community **llama.cpp / Ollama** pack derived from the Ollama library [nemotron-3-super](https://ollama.com/library/nemotron-3-super) **Q4_K_M** weights (NVIDIA Nemotron 3 Super, ~120B total / ~12B active MoE).

This is **not** an official NVIDIA Transformers / NIM / NVFP4 safetensors checkpoint. Official weights live under [nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4) (and sibling FP8/BF16 repos). This repo is a **Q4_K_M GGUF** with **metadata-only** YaRN so runners that honor `n_ctx_train` accept **1,048,576** tokens.

| | Official NVIDIA Hub (e.g. NVFP4) | This repo |
|--|--|--|
| Format | Safetensors / NVIDIA stacks | GGUF (`nemotron_h_moe`) |
| Quant | NVFP4 / FP8 / BF16 | **Q4_K_M** (from Ollama library) |
| Context in file | Up to 1M (NVIDIA card) | Library GGUF advertised **262144** → we set **1048576** + YaRN KV |
| Typical serve | NIM, vLLM, Transformers | llama.cpp, Ollama |

License: **NVIDIA Nemotron Open Model License** (same family as the official Super card). The [`LICENSE`](LICENSE) file is the NVIDIA Open Model License Agreement text (Last Modified: October 24, 2025). See also [license link](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-nemotron-open-model-license/).

---

## What you get

| File | Role | Size |
|------|------|------|
| `Nemotron-3-Super-Q4_K_M-yarn1m-00001-of-00003.gguf` | Shard 1/3 (Hub <50 GB limit) | ~39 GiB |
| `Nemotron-3-Super-Q4_K_M-yarn1m-00002-of-00003.gguf` | Shard 2/3 | ~39 GiB |
| `Nemotron-3-Super-Q4_K_M-yarn1m-00003-of-00003.gguf` | Shard 3/3 | ~7 GiB |
| `Modelfile` | Ollama example (`FROM` first shard, `num_ctx` 1M, `RENDERER`/`PARSER` `nemotron-3-nano`) | tiny |

The single ~87 GiB file exceeds Hugging Face’s **50 GB per-file** limit, so this pack is **`llama-gguf-split --split-max-size 40G`**. Ollama / llama.cpp load the set from the first shard path.

**1M context is not extra trained weights.** Tensors are unchanged from the Ollama Q4_K_M blob. Only GGUF key-values were rewritten:

| Key | Value |
|-----|--------|
| `nemotron_h_moe.context_length` | `1048576` |
| `nemotron_h_moe.rope.scaling.type` | `yarn` |
| `nemotron_h_moe.rope.scaling.factor` | `4.0` |
| `nemotron_h_moe.rope.scaling.original_context_length` | `262144` |

Sparse attention keeps 1M KV cheap on unified-memory hosts (~**8 GB** KV class + ~81–87 GB weights ≈ **~94–95 GB** resident on a DGX Spark when fully allocated).

---

## Why this pack exists

NVIDIA’s Super card targets **up to 1M** context. The published Ollama library GGUF only advertised **262144**, and stock Ollama **caps** `num_ctx` at GGUF `n_ctx_train`. Setting `PARAMETER num_ctx 1048576` alone left `ollama show` at 262K.

This pack patches KV (same pattern as our [Qwen3.8 NVFP4+VL YaRN pack](https://huggingface.co/mixbits/Qwen3.8-27B-NVFP4-MTP-VL-GGUF)) so Ollama / llama.cpp will allocate the full window.

---

## Validated on DGX Spark (GB10, 128 GB unified)

Host: Docker `ollama/ollama` (~0.32.x), tag **`nemotron-3-super:1m`**, Spark `OLLAMA_CONTEXT_LENGTH` left at **65536** (fallback only — Modelfile pins 1M).

| Check | Result |
|-------|--------|
| `ollama show` context length | **1048576** |
| Default load (`ollama ps` CONTEXT) | **1048576** / **~94 GB** |
| Needle ~8k / 64k / 256k / **512k** | **PASS** |
| Single-shot needle ~**980k** tokens | **PASS** (~42 min, ~388 t/s prompt) |

Secret used for the full-depth run: `SECRET_CODE_NEMO_SUPER_1M_FULL_9917` (recalled correctly). Global env was **not** raised to 1M; precedence is request → Modelfile → `OLLAMA_CONTEXT_LENGTH`.

---

## Use

Needs a recent Ollama / llama.cpp build that understands **`nemotron_h_moe`**. Keep **`RENDERER`/`PARSER` `nemotron-3-nano`** for tools + thinking.

### Ollama (after `hf download`)

```bash
hf download mixbits/Nemotron-3-Super-Q4_K_M-YaRN-1M-GGUF --local-dir ./nemotron-3-super-yarn1m
cd ./nemotron-3-super-yarn1m
# Modelfile FROM points at shard 00001-of-00003; keep all three shards together
ollama create nemotron-3-super:1m -f Modelfile
ollama run nemotron-3-super:1m
```

### API smoke

```bash
curl -s localhost:11434/api/chat -d '{
  "model": "nemotron-3-super:1m",
  "stream": false,
  "think": false,
  "options": {"num_predict": 64},
  "messages": [{"role": "user", "content": "Hello"}]
}'
```

Leaving Open WebUI `num_ctx` as **Default** uses Modelfile **1M**. Do not raise a global Ollama `OLLAMA_CONTEXT_LENGTH` to 1M just for this tag — that hits every unpinned model.

### Recreate YaRN GGUF from stock Q4

If you still have the library blob (or re-`ollama pull nemotron-3-super:latest`):

```bash
python scripts/nemotron-super-yarn-1m-rewrite.py \
  /path/to/stock-q4.gguf \
  ./Nemotron-3-Super-Q4_K_M-yarn1m.gguf
```

---

## Honesty checklist

- Do **not** claim official NVIDIA NVFP4 / FP8 / BF16 safetensors.
- Do **not** claim we re-quantized from Hub weights; language tensors came from the **Ollama library Q4_K_M** blob, then KV rewrite only.
- Do **not** claim Modelfile-only `num_ctx` is enough without GGUF `n_ctx_train` ≥ 1M.
- **Do** claim DGX Spark validation: 1M KV alloc + needle through ~512k and a single-shot ~980k recall **PASS** (2026-08-23).

---

## Attribution

- Model family: NVIDIA Nemotron 3 Super ([tech report](https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Super-Technical-Report.pdf), [Hub NVFP4](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4))
- Q4_K_M distribution: [ollama.com/library/nemotron-3-super](https://ollama.com/library/nemotron-3-super)
- YaRN 1M GGUF rewrite + Spark validation: mixbits (this repo)