Reading path
I just bought a DGX Spark, where do I start?
The unboxing-to-first-token order. Five articles, read top to bottom, and you have a working self-hosted model on the box.
5 articles, in reading order
- Self-Hosted AI: Start Here
Start here. The map of everything: what runs where, in what order, and why.
- Self-Host Mistral Small 4 with SGLang on NVIDIA DGX Spark (GB10): What Actually Works
First model up. The SGLang inference setup on GB10, including the kernel constraints that bite on day one.
- ComfyUI plus FLUX.1-schnell on DGX Spark: Per-Style Visual Vocabularies
Image generation on the same box, sharing the unified memory pool one service at a time.
- Two Days From Localhost to Production: Building a Hybrid Sovereign AI Site
What a full stack looks like once it is running, end to end, in two days of real work.
- Spark Arena Rank 4 Made Me Add Qwen3.6 to My DGX Spark
Which model to actually run, and why the answer keeps moving as better options ship for GB10.