Cohere North Mini Code — 3B Active Params Crushes Hard LeetCode Problems

About this lesson

Cohere North Mini Code is a 30B MoE model with only 3B active parameters — and it just solved 4 hard LeetCode problems live on a RunPod H100 via vLLM. Here's the full breakdown: architecture, benchmarks, speed, and live demo. North Mini Code uses a Mixture-of-Experts architecture (FP8 quantized) that activates just 3B params per token — making it fast and memory-efficient without sacrificing coding accuracy. Tested on LRU Cache, Trapping Rain Water, Course Schedule II, and Serialize Binary Tree. 🔧 Stack: - Model: CohereLabs/North-Mini-Code-1.0-fp8 (HuggingFace) - Inference: vLLM on RunPod H100 - Blog: https://cohere.com/blog/north-mini-code ⏱️ Chapters: 0:00 — What is North Mini Code? 0:17 — MoE Architecture (30B total / 3B active) 0:54 — Benchmarks vs other coding models 1:19 — Speed & token throughput 1:40 — Live demo intro (RunPod + vLLM) 1:55 — Problem 1: LRU Cache (Hard) 2:20 — Problem 2: Trapping Rain Water (Hard) 2:44 — Problem 3: Course Schedule II (Hard) 3:10 — Problem 4: Serialize Binary Tree (Hard) 3:49 — Verdict #NorthMiniCode #Cohere #LocalLLM 🔗 Connect with me: Ko-fi → https://ko-fi.com/promptengineer Patreon → https://www.patreon.com/PromptEngineer975 Book a Call → https://calendly.com/prompt-engineer48/call GitHub → https://github.com/PromptEngineer48 Twitter/X → https://twitter.com/prompt48

Original Description