**This is the 2.8T-parameter Kimi K3 running at ~30 tok/s on 16× NVIDIA DGX Spark.**
This is a dev post classified by Jev as AI dev tools (a tool drop), kept by the Dev Radar because it carries real work, not commentary.
**This is the 2.8T-parameter Kimi K3 running at ~30 tok/s on 16× NVIDIA DGX Spark.** Not a synthetic “it boots” test. 👇 **Actually generating. Usable speed.** A few weeks ago, the question was: *Can a model this big even run properly on a cluster of Sparks?* Then: *Can we make it fast enough to actually use? What about for agentic frameworks?* I think we're well past that question now. With the new **V5**, we're seeing roughly another **20% speed-up**, plus a big improvement in concurrency: ⚡ **C1: 29.8 tok/s** ⚡ **C2: 42 tok/s aggregate** ⚡ **C4: 58 tok/s aggregate** ⚡ **C8: 87.1 tok
Posted by llmwish.open (1.6k followers) 1 days ago · 172 likes · 432.2k views · view the original post on X. Kept by the Dev Radar as AI dev tools. Tools mentioned: gb10-vllm, NVIDIA Developer Forums.
More dev work like this
- If you're doing text rendering with @Alibaba_Qwen Qwen-Image-2.1 then this one is for you. — @ItsmeAjayKV
- 🔥最新：DeepSeek Harness v0.1.7-alpha.1 已更新！ — @NFT_Chen
- Grok 4.7 High vs Xiaomi MiMo-V2.6 ! — @YouWareAI
- 🚨突发！阶跃星辰 Step Code v0.1.0 正式开源！ — @NFT_Chen
- Built a Rocket Launch Scene in Three.js with Grok 4.7 and compared it with Step 5 Preview — @karankendre
- Tencent ARC Lab just released GameHorizon Suite — @HuggingPapers
- The Dot platform has processed 6.18B tokens this week, representing 34.33% of our… — @usedotai
- One bad agent command can erase a working tree in seconds. — @DanKornas
Every post is read and classified by Jev (TypeSafe): what it is, which market it belongs to, and whether the link is a real tool. 20.7k posts from 4.9k X accounts over the last 21 days, 2.4k tools, 12 markets. Collected every 5 minutes, fully re-ranked every hour — last update 2026-09-22 09:16 UTC. Full method.