← Model Cards

GB10SOR LANES · 2026-09-21

Solo

1 Spark · 7 models · 9 qualified recipes

NixOS

Choose a recipe, stage its weights, and run the command from the matching GB10SOR checkout. Multi-Spark lanes also need the private local cluster configuration described in the repository. Setup guide →

Use the repository quickstart for DGX OS or NixOS. Support still depends on the exact model recipe and a prepared host.

MiniMaxAI

MiniMax-Music3

MiniMaxAI/MiniMax-Music3

NixOSOne-command Nix ready
Recipe options
Engine
SGLang Omni 0.1.3
Served context
Not recorded
Request limit
Not recorded
Parallelism
TP1 / PP1
Checkpoint weights
53.4 GiB
Recipe YAML

No YAML linked to this recipe.

Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: MiniMax model team; Mikehutu; ComfyUI contributors

Checkpoint revision: fbdf52fbaaca799592917417eb05f1899f1255ec

Recipe SHA-256: Not recorded

Exact GB10 Music3/H3 toolkit ↗exact model card ↗
nvidia

NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4

nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-NVFP4

NixOSOne-command Nix ready
Recipe options
Engine
vLLM 0.27.1
Served context
65,536 tokens
Request limit
4
Parallelism
TP1 / PP1
Checkpoint weights
20.1 GiB
Recipe YAML
Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: NVIDIA Nemotron/TensorRT-LLM teams; sojufx; Atlas Inference; vLLM/SGLang upstream

Checkpoint revision: e8f3c7c4de75ad84fe1bcef95d38eca76214480b

Recipe SHA-256: a98f767ea781f102915e132918063dcefda6b3d225b6dd3b8e441f7952d93877

Official NVIDIA multi-engine cookbook ↗exact hardened vLLM Solo recipe ↗
nvidia

Qwen3.8-Flash-Next-NVFP4

nvidia/Qwen3.8-Flash-Next-NVFP4

NixOSOne-command Nix ready
Recipe options
Engine
vLLM 0.28.1rc1.dev388+g8a728663c
Served context
262,144 tokens
Request limit
6
Parallelism
TP1 / PP1
Checkpoint weights
123.6 GiB
Recipe YAML
Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: NVIDIA model team; Tony DiAngelo (tonyd2wild); vLLM upstream

Checkpoint revision: fab0aecb760cec45227f6656abcaafa11abca87a

Recipe SHA-256: 2557c5bb7ae30e0c72571b630e1831d90126d9c30a6e8192ec536e0acc697f47

Exact NVIDIA-quant Spark recipe ↗
OpenMed

privacy-filter-nemotron-v2

OpenMed/privacy-filter-nemotron-v2

NixOSOne-command Nix ready
Recipe options
Engine
Transformers 5.15.0
Served context
Not recorded
Request limit
Not recorded
Parallelism
TP1 / PP1
Checkpoint weights
2.6 GiB
Recipe YAML

No YAML linked to this recipe.

Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: OpenMed model team; NVIDIA Nemotron upstream; Transformers/vLLM contributors

Checkpoint revision: 968247329d18998cc7d5338b941b52f1b2a9abd9

Recipe SHA-256: Not recorded

Exact model card ↗
RadixArk

Qwen3.8-27B-NVFP4-BF16-LMHead

RadixArk/Qwen3.8-27B-NVFP4-BF16-LMHead

NixOSOne-command Nix ready
Recipe options
Engine
SGLang 0.0.0.dev1+g5f55db35e
Served context
65,536 tokens
Request limit
Not recorded
Parallelism
TP1 / PP1
Checkpoint weights
22.1 GiB
Recipe YAML
Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: RadixArk; MiaAI-Lab; SGLang cookbook maintainers; Qwen and z-lab DFlash contributors

Checkpoint revision: 009632fef96dd349150baa780c984e62e70e91fe

Recipe SHA-256: 54e71542d85f5ee43dea7250ad238f6334a9ba45695a8346a65d97408abe0c27

Exact maintained GB10 launcher ↗official SGLang cookbook ↗
RadixArk

Qwen3.8-Flash-Next-NVFP4

RadixArk/Qwen3.8-Flash-Next-NVFP4

NixOSOne-command Nix ready
Recipe options
Engine
vLLM 0.1.dev20073+g8e685d198
Served context
262,144 tokens
Request limit
2
Parallelism
TP1 / PP1
Checkpoint weights
125.9 GiB
Recipe YAML
Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: RadixArk; sfxnz; MiaAI-Lab; SGLang/vLLM upstream

Checkpoint revision: 7b719225242aacd3dbd3f9407468c2ee9a9d2594

Recipe SHA-256: 58a9221794d70e57909e10f7632221d7f62bcad4173aa0d46f5ec73cab2bd55e

Exact TP2 recipe ↗SGLang family recipe ↗
RedHatAI

Muse-Glimmer-30B-NVFP4

RedHatAI/Muse-Glimmer-30B-NVFP4

NixOSOne-command Nix ready
Recipe options
Engine
vLLM 0.26.1rc1.dev608+g99a10304d
Served context
65,536 tokens
Request limit
4
Parallelism
TP1 / PP1
Checkpoint weights
21.8 GiB
Recipe YAML
Recipe & author credits

Use the exact checkpoint revision and deployment profile. Combining recipes is a planning choice; it does not establish combined compatibility.

Contributors: Red Hat AI model/quantization team; MiaAI-Lab; chishiki37; vLLM upstream; GB10SOR local qualification

Checkpoint revision: d5109a1d187c27bd1734e81844e71aa4d964e66a

Recipe SHA-256: 66d517b8cb898a5cada8beb940d794d013c1af05b6f8944be73d4f04a5a68b2e

Exact NVFP4 GB10 recipe ↗