Written by Jakub Rusinowski · Last updated July 21, 2026
Best all-round pick: GPT-oss 120B
The NVIDIA DGX Spark has 128 GB of VRAM, of which about 128 GB is available to a model. The largest model it can hold is Devstral-2 123B (123B, 75.1 GB at Q4_K_M).
| Model | Score | Memory at Q4_K_M | Context | Licence |
|---|---|---|---|---|
| GPT-oss 120B | 96 | 73.3 GB | 125K | Apache-2.0 |
| Nemotron 70B Instruct | 94.7 | 43.4 GB | 125K | Llama Community |
| Qwen 3.5 122B-A10B (MoE) | 94.2 | 74.5 GB | 125K | Apache 2.0 |
| Qwen 3.5 122B-A10B | 92.5 | 74.5 GB | 128K | Apache-2.0 |
| GLM-5.1 72B | 91.2 | 44.3 GB | 125K | MIT |
| Command R+ (104B) | 91.2 | 63.6 GB | 125K | CC-BY-NC |
| Workload | Recommended on this GPU |
|---|---|
| Coding | Devstral-2 123B (98.5) Qwen 3.5 122B-A10B (MoE) (96.5) Qwen3-Coder 80B-A3B (MoE) (96.2) |
| General assistant | GPT-oss 120B (96) Nemotron 70B Instruct (94.7) Qwen 3.5 122B-A10B (MoE) (94.2) |
| Reasoning | GLM-5.1 72B (98.4) Qwen 3.5 122B-A10B (MoE) (97) Gemma 4 27B ⭐ (95.9) |
| RAG | Llama 4 Scout 17B (93.6) Gemma 4 31B (92.8) Llama 4.5 Scout (92.3) |
| Agents | Ternary Bonsai 27B (89) Qwen 3.5 122B-A10B (MoE) (87.5) Nemotron 3 Super 120B-A12B (86.6) |
| Vision | Qwen 3.5 72B (96.9) Llama 3.2 90B Vision Instruct (95.2) Mistral Small 4 119B-A6.5B (93.1) |
GPT-oss 120B is the strongest all-round pick that fits its 128 GB.
Devstral-2 123B — 123B parameters, needing 75.1 GB at Q4_K_M.
About 128 GB of its 128 GB, before the desktop and runtime overhead are accounted for.