744B on a Laptop? colibrì Actually Runs It (25GB RAM)
colibrì runs a 744-billion-parameter GLM-5.2 frontier model on just 25 GB of RAM — no GPU, no cloud, pure C streaming from disk. In this deep dive, we break down how the tiny C engine makes the impossible possible. Mixture-of-experts architecture means only ~40B parameters fire per token. The resident dense core (~9.9 GB) stays in RAM while the remaining 370 GB of routed experts stream off your SSD on demand. Multi-head latent attention compresses the KV cache by 57×. Speculative decoding with an int8 MTP head pushes useful latency down. And the engine literally learns you — recording your expert usage to pin your hottest paths into RAM. We cover the architecture, the disk-to-RAM pipeline, MLA compressed KV, MTP speculative decoding, the learning cache, real community benchmarks, and the honest truth about what colibrì is: a 744B model answering where nothing else can. Slowly, but it runs. ⏱ CHAPTERS 0:00 — Intro 0:04 — The Impossible 0:20 — Why Clusters 1:31 — Architecture Deep Dive 3:50 — It Learns You 4:58 — How To Run 7:00 — The Honest Take *Repo:* https://github.com/JustVugg/colibri *Pre-converted int4 model:* https://huggingface.co/jlnsrk/GLM-5.2... *Int8 MTP heads:* https://huggingface.co/mateogrgic/GLM... colibrì v1.0 — Apache 2.0 engine, MIT weights. Pure C, no BLAS, no Python at runtime, no GPU required. Compile with gcc and OpenMP, set COLI_MODEL, run `coli chat`. Under a minute to a prompt. If you run it, post your benchmark numbers to the repo — the community is building the real-world speed database as we speak. #OpenSource #AI #colibri #GLM5 #LocalLLM #EdgeInference #MachineLearning #DeveloperTools #opensource #aiterminal #LLM

The Local AI Hardware Mistake Everyone Makes

I Tested the Cheapest Path to 96GB of VRAM

How To: I Turned 96GB System RAM Into “VRAM” on AMD iGPU (Linux How-To)

The Hard Drive Cartel | Criminal Conspiracy & Price Fixing

Agents A1 Benchmarked vs Qwen 35B - 16GB Local LLM setup

Mojo + Vulkan is INSANE: Run Local AI on ANY GPU (Goodbye CUDA)

I was right again

We Just Hit the Local LLM Tipping Point (Colibri)

NVIDIA didn't want me to do this

Ornith 35B Benchmarked vs Qwen 35B - 16GB Local LLM setup

China Is About To Pop The AI Bubble

DGX Spark's Fatal Flaw Gets Fixed—Here's What NVIDIA Missed

I Hacked This Temu Router. What I Found Should Be Illegal.

Everything That Actually Matters for Local AI

The Best Local Agentic Coding Workflow (Complete Guide)

This Tiny 1.9MB Tool Is Destroying Microsoft’s Windows 11 Plans

Casey Muratori – The Big OOPs: Anatomy of a Thirty-five-year Mistake – BSC 2025

KEEP AI LOCAL! Explaining Agentic AI and The Loop

AMAZON's New 5TB/S MONSTER Chip Just Made Google & Nvidia's AI GPUs Look Like PAPER WEIGHTS!

