Model library
Open models, with the GPU to run them.
Each model below names the GPU we recommend for it and opens the console with that GPU and model already picked. One-click presets start serving an OpenAI-compatible API on your own machine; the others you run yourself with the linked guide. SSH, the browser terminal and the rest of the machine stay yours either way.
Featured models
One-click presets: the weights are already on the machine, pinned to a fixed revision.
Qwen3 Coder 30B
A coding model for agents and editors, with tool calls turned on.
Qwen3.6 35B-A3B
A fast mixture-of-experts model for chat and agents, also on the pay-per-token API.
GPT-OSS 120B
OpenAI's open-weight 120B model, quantized to fit one DGX Spark.
Qwen3.8 Flash Next
A large Qwen3.8 mixture-of-experts model, quantized to fit one DGX Spark.
GLM-5.3-Flash
A 320B mixture-of-experts model at a sub-2-bit quant, for the longest context on one Spark.
ComfyUI
The node editor for image and video workflows, in your browser. Bring your own checkpoints.
All models
Recommended GPU, rate and availability are read live. A one-click preset deploys the model; the others deploy a base image and link the setup guide.
| Model | Engine | Size | Recommended GPU | View or deploy |
|---|---|---|---|---|
| vLLM, llama.cppOne-click preset | AWQ 4-bit · 5 configurations | RTX 5090$0.69 / hour · Check availability | ||
| llama.cpp, vLLMOne-click preset | Q8_0 GGUF · 5 configurations | DGX Spark$0.79 / hour · Check availability | ||
| llama.cpp, vLLMOne-click preset | UD-Q4_K_XL GGUF · 2 configurations | DGX Spark$0.79 / hour · Check availability | ||
| llama.cpp, vLLMOne-click preset | UD-Q3_K_XL GGUF · 4 configurations | DGX Spark$0.79 / hour · Check availability | ||
| llama.cpp, vLLMOne-click preset | UD-IQ1_S GGUF · 2 configurations | DGX Spark$0.79 / hour · Check availability | ||
| ComfyUIOne-click preset | Bring your own checkpoints | DGX Spark$0.79 / hour · Check availability | ||
| vLLM, llama.cppRun it yourself | NVFP4 · 4 configurations | DGX Spark$0.79 / hour · Check availability | ||
| vLLM, llama.cppRun it yourself | NVFP4, GGUF or BF16 · 5 configurations | DGX Spark$0.79 / hour · Check availability | ||
| vLLM, llama.cppRun it yourself | NVFP4 · 2 configurations | DGX Spark$0.79 / hour · Check availability | ||
| vLLM, llama.cppRun it yourself | NVFP4 · 2 configurations | DGX Spark$0.79 / hour · Check availability | ||
| llama.cpp, vLLMRun it yourself | MXFP4 · 3 configurations | DGX Spark$0.79 / hour · Check availability | ||
| vLLM, llama.cppRun it yourself | AWQ 4-bit · 2 configurations | DGX Spark$0.79 / hour · Check availability | ||
| vLLM, llama.cppRun it yourself | NVFP4 · 3 configurations | DGX Spark$0.79 / hour · Check availability | ||
| vLLMRun it yourself | FP8 on two Sparks, 2-bit builds on one | 2 × DGX Spark$1.79 / hour · Check availability | ||
| llama.cpp, vLLMRun it yourself | Q4_K_XL GGUF | 2 × DGX Spark$1.79 / hour · Check availability |
No model in this category yet.
Model list checked 21 November 2026. Compare every GPU · Billing and stopped instances · Which models fit in 128 GB
Deploy your own
Not in the list? Every GPU you rent is a whole machine with root in your own container, so any open model that fits its memory runs on it.
Any model with vLLM
Deploy a base image, download the weights from Hugging Face and serve them with vLLM.
vLLM guide →Any GGUF with llama.cpp
Pick the quant that fits your GPU's memory and serve it with llama-server.
llama.cpp guide →Your checkpoints in ComfyUI
Start the ComfyUI preset and add the checkpoints and LoRAs you want.
ComfyUI →Pay per token instead
Call an open model through the inference API without renting a GPU.
Inference API →Which models fit in 128 GB · Quantization explained · Choosing an inference engine