Back AI Curator

LocalLLaMA Reddit2d ago

Vercel CEO: "Almost shocked" by how good GLM-5.2 is at coding

AI is generating summary...

Comments

No comments yet

Be the first to comment

Related Articles

I forked ik_llama.cpp and added a "--numa mirror" mode to m…

I pretrained and post trained a 500M parameter LLM and 330M…

2× Radeon R9700 — Qwen 3.6 27B Q8 MTP on llama.cpp

Tokenomics

ROCm vs Vulkan vs vLLM on Dual R9700's

Rollin' MiMo-2.5 on two Halo Strixeses

8-16 MI50s Minimax M3 @19 tps TG (peak)

Why is AutoRound being slept on so hard?

Watch local LLMs escape the rooms you design

Gemma 4 QAT seems to respond significantly better to KV cac…

AI Curator

Your AI news assistant

Ask me anything about AI

I can help you understand AI news, trends, and technologies