NVIDIA - desktop AI system - verified
Local LLMs that run on NVIDIA DGX Spark (up to 128 GB)
By this directory's rule of thumb, 42 of 63 open-weight models fit in the 128 GB of unified memory on the DGX Spark in its largest configuration. The largest by total parameters is Mistral Medium 3.5 128B (127.7B), whose smallest non-MLX download is 74.9 GB, the Q4_K_M GGUF build from unsloth/Mistral-Medium-3.5-128B-GGUF.
- Models that fit
- 42 of 63
- Memory
- Up to 128 GB
- Memory type
- LPDDR5X unified
- Bandwidth
- 273 GB/s
Spec sheet
- Memory
- Up to 128 GB
- Configurations
- 64 or 128 GB
- Memory type
- LPDDR5X unified
- Unified memory
- Yes, shared by CPU and GPU
- Bandwidth
- 273 GB/s
Offered with 64 GB or 128 GB of coherent unified system memory on a 256-bit interface, powered by the GB10 Superchip.
Source: NVIDIA spec page (nvidia.com)
42 models that fit in 128 GB
Total params, largest first| Model | Total params | Smallest download | Quant | Commercial |
|---|---|---|---|---|
| Mistral Medium 3.5 128BMistral AI | 127.7B | 74.9 GB | Q4_K_MGGUF | Commercial use: Conditions |
| Devstral 2 123B Instruct 2512Mistral AI | 125B | 74.9 GB | Q4_K_MGGUF | Commercial use: Conditions |
| NVIDIA Nemotron 3.5 Super VL 120B-A12BNVIDIA | 124.5B12B active | 71.8 GB | Q4_0GGUF | Commercial use: Yes |
| Mistral Small 4 119B (2603)Mistral AI | 119.4B6.5B active | 73.8 GB | UD-Q4_K_MGGUF | Commercial use: Yes |
| gpt-oss-120bOpenAI | 116.8B5.1B active | 62.8 GB | Q4_K_MGGUF | Commercial use: Yes |
| Cohere Command A (03-2025)Cohere | 111.1B | 67.1 GB | Q4_K_MGGUF | Commercial use: No |
| GLM-4.5-AirZ.ai | 110.5B12B active | 73.0 GB | Q4_K_MGGUF | Commercial use: Yes |
| Llama 4 Scout 17B-16E InstructMeta | 108.6B17B active | 65.4 GB | Q4_K_MGGUF | Commercial use: Conditions |
| GLM-4.6VZ.ai | 107.7B | 70.6 GB | Q4_K_MGGUF | Commercial use: Yes |
| Qwen3-Coder-NextAlibaba Cloud (Qwen) | 79.7B3B active | 48.4 GB | Q4_K_MGGUF | Commercial use: Yes |
| NVIDIA Nemotron 3.5 Bolt 74B-A9BNVIDIA | 74.9B9B active | 22.0 GB | 24GB (native 2-bit)GGUF | Commercial use: Yes |
| AI21 Jamba2 MiniAI21 Labs | 51.6B12B active | 31.2 GB | Q4_K_MGGUF | Commercial use: Yes |
| Qwen3.6-35B-A3BAlibaba Cloud (Qwen) | 36B3B active | 22.1 GB | UD-Q4_K_MGGUF | Commercial use: Yes |
| Cohere Aya Expanse 32BCohere | 32.3B | 19.8 GB | Q4_K_MGGUF | Commercial use: No |
| Ai2 Olmo 3.1 32B ThinkAi2 | 32.2B | 19.5 GB | Q4_K_MGGUF | Commercial use: Yes |
| NVIDIA Nemotron 3.5 Lightning 30B-A3BNVIDIA | 31.6B3B active | 18.9 GB | Q4_0GGUF | Commercial use: Yes |
| Gemma 4 31BGoogle | 31.3B | 17.7 GB | Q4_0 (QAT)GGUF | Commercial use: Yes |
| IBM Granite 4.2 30BIBM | 29.3B | 17.7 GB | Q4_K_MGGUF | Commercial use: Yes |
| Qwen3.8-27BAlibaba Cloud (Qwen) | 27.8B | 16.5 GB | UD-Q4_K_MGGUF | Commercial use: Yes |
| Gemma 4 26B A4BGoogle | 25.8B3.8B active | 14.4 GB | Q4_0 (QAT)GGUF | Commercial use: Yes |
| Devstral Small 2 24B Instruct 2512Mistral AI | 24B | 14.3 GB | Q4_K_MGGUF | Commercial use: Yes |
| Liquid AI LFM2.5-24B-A2BLiquid AI | 24B2.5B active | 14.5 GB | Q4_K_MGGUF | Commercial use: Conditions |
| gpt-oss-20bOpenAI | 20.9B3.6B active | 11.6 GB | Q4_K_MGGUF | Commercial use: Yes |
| Phi-4-reasoning-vision-15BMicrosoft | 15.1B | 30.2 GB | BF16Official weights | Commercial use: Yes |
| Phi-4Microsoft | 14.7B | 9.1 GB | Q4_KGGUF | Commercial use: Yes |
| Phi-4-reasoning-plusMicrosoft | 14.7B | 9.1 GB | Q4_K_MGGUF | Commercial use: Yes |
| Ministral 3 14B Instruct 2512Mistral AI | 13.9B | 8.2 GB | Q4_K_MGGUF | Commercial use: Yes |
| Gemma 4 12BGoogle | 12B | 7.0 GB | Q4_0 (QAT)GGUF | Commercial use: Yes |
| Qwen3.5-9BAlibaba Cloud (Qwen) | 9.7B | 5.7 GB | Q4_K_MGGUF | Commercial use: Yes |
| Ministral 3 8B Instruct 2512Mistral AI | 8.9B | 5.2 GB | Q4_K_MGGUF | Commercial use: Yes |
| IBM Granite 4.2 8BIBM | 8.8B | 5.3 GB | Q4_K_MGGUF | Commercial use: Yes |
| Cohere Command R7B (12-2024)Cohere | 8B | 5.1 GB | Q4_K_MGGUF | Commercial use: No |
| Gemma 4 E4BGoogle | 8B | 5.0 GB | Q4_K_MGGUF | Commercial use: Yes |
| Ai2 Olmo Hybrid Instruct DPO 7BAi2 | 7.4B | 14.9 GB | BF16Official weights | Commercial use: Yes |
| Gemma 4 E2BGoogle | 5.1B | 3.1 GB | Q4_K_MGGUF | Commercial use: Yes |
| Qwen3.5-4BAlibaba Cloud (Qwen) | 4.7B | 2.7 GB | Q4_K_MGGUF | Commercial use: Yes |
| Ministral 3 3B Instruct 2512Mistral AI | 3.8B | 2.1 GB | Q4_K_MGGUF | Commercial use: Yes |
| Phi-4-mini-instructMicrosoft | 3.8B | 2.5 GB | Q4_K_MGGUF | Commercial use: Yes |
| Phi-4-mini-reasoningMicrosoft | 3.8B | 2.5 GB | Q4_K_MGGUF | Commercial use: Yes |
| Liquid AI LFM2.5-2.6BLiquid AI | 2.7B | 1.7 GB | Q4_K_MGGUF | Commercial use: Conditions |
| Qwen3.5-2BAlibaba Cloud (Qwen) | 2.3B | 1.3 GB | Q4_K_MGGUF | Commercial use: Yes |
| Qwen3.5-0.8BAlibaba Cloud (Qwen) | 0.9B | 0.5 GB | Q4_K_MGGUF | Commercial use: Yes |
The rest of the 63 need more than 128 GB. Browse every local model.
By memory configuration
NVIDIA sells the DGX Spark with 64 or 128 GB. The same rule applied to each configuration, and the models each step up adds.
| Configuration | Models that fit | Largest that fits | New at this size |
|---|---|---|---|
| 64 GB | 33 / 63 | Qwen3-Coder-Next | Base configuration: 33 models |
| 128 GB | 42 / 63 | Mistral Medium 3.5 128B | Mistral Medium 3.5 128B, Devstral 2 123B Instruct 2512, NVIDIA Nemotron 3.5 Super VL 120B-A12B, Mistral Small 4 119B (2603), gpt-oss-120b, Cohere Command A (03-2025), GLM-4.5-Air, Llama 4 Scout 17B-16E Instruct, GLM-4.6V |
Other devices with 128 GB
Same memory size. Each shows how many models fit under its own rule (Macs also count MLX builds), and the memory type and bandwidth differ.
Compare all 33 GPUs and Macs by memory, bandwidth and how many models fit.
FAQ
How many local LLMs can the DGX Spark run?
42 of the 63 open-weight models in this directory fit in its largest configuration, 128 GB, by this site's rule of thumb that the smallest non-MLX download is at most 85% of the memory. By configuration: 33 at 64 GB and 42 at 128 GB. MLX builds do not count here: they run on Apple silicon only.
Can the DGX Spark run Mistral Medium 3.5 128B?
Yes, by this directory's rule of thumb: its smallest non-MLX download, 74.9 GB, is at most 85% of 128 GB (108.8 GB), the Q4_K_M GGUF build from unsloth/Mistral-Medium-3.5-128B-GGUF. That leaves room for the context and the operating system. It is a memory check, not a speed estimate.
Can the DGX Spark run NVIDIA Nemotron 3.5 Super VL 120B-A12B?
Yes, by this directory's rule of thumb: its smallest non-MLX download, 71.8 GB, is at most 85% of 128 GB (108.8 GB), the Q4_0 GGUF build from ggml-org/NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-GGUF. That leaves room for the context and the operating system. It is a memory check, not a speed estimate.
Can the DGX Spark run Step 3.7 Flash?
Not by this directory's rule of thumb: its smallest non-MLX download, 111.5 GB, is more than 85% of 128 GB (108.8 GB). It needs the 256 GB tier or more: https://www.developersdigest.tech/local-models/fits/256.
Is the DGX Spark's memory shared with the operating system?
Yes. NVIDIA lists LPDDR5X unified memory shared by the CPU and GPU, so the operating system and your apps use the same pool as the model. The 15% headroom in this directory's rule is there for that and for the context.
What memory bandwidth does the DGX Spark have?
NVIDIA states 273 GB/s of memory bandwidth (LPDDR5X unified). From the spec notes: Offered with 64 GB or 128 GB of coherent unified system memory on a 256-bit interface, powered by the GB10 Superchip. This page does not turn bandwidth into a speed estimate.
Newsletter
New open weights, every week
One email per week. New model releases, open-weight drops and what they mean for what you build.
Keep exploring
Best local AI models
Our curated picks by hardware tier, with benchmarks and the reasoning.
Model API pricing and specs
First-party prices per 1M tokens when you would rather call an API.
AI benchmark leaderboards
Sourced scores, for when you need to compare quality.
Setup guides
Point your coding tools at the model you picked.