DeepSeek v4 Flash 0731 on 2x DGX Sparks

All projects

Apps & Tools

DeepSeek v4 Flash 0731 on 2x DGX Sparks

Optimized 1M-context deployment recipe for serving DeepSeek-V4-Flash on DGX clusters.

byMia'a AI Lab

Open source

DeepSeek v4 Flash 0731 on 2x DGX Sparks media is blocked

Allow external media to connect to the provider and play this content.

Description

A high-performance deployment recipe for serving DeepSeek-V4-Flash-0731 with vLLM and DSpark speculative decoding. It features a validated 1M-token context profile using the experimental nvfp4_ds_mla KV-cache path on dual DGX Spark nodes.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: Impressive technical implementation proving that 1M-token context windows can be served with high throughput on…
Attributed to: Mia'a AI Lab@MiaAI-Lab
Built with: Muse Spark 1.2 DeepSeek-V4-Flash-0731 Qwen3.7 Flash 3.5 Flash Cyber 3.6 Flash Spark 1.1 Step 3.7 Flash Gemini 3.5 Flash 2.5 Flash
Links: View tweet
Tags: #llm #vllm #infrastructure #deployment

Discovered: X
First posted: X · Aug 4, 2026
Listed: Added 2h ago
Build: Strong
Models: Strong