INTELLECT-3: A 100B+ MoE Trained with Large-Scale RL on GLM 4.5 AIR @fahdmirza

INTELLECT-3: A 100B+ MoE Trained with Large-Scale RL on GLM 4.5 AIR @fahdmirza

Updated September 2026, 6 days ago

INTELLECT-3: A 100B+ MoE Trained with Large-Scale RL on GLM 4.5 AIR

Ornith-1.5-9B: Great on Paper, Struggled in My Tests Locally

Gemma4 12B in Quantization-Aware Training (QAT) with Ollama - Full Testing

NVIDIA Nemotron 3 Embed 1B Cuts Agent Token Costs by 31%

Thomson Reuters Built Their Own AI Lawyer: Run Thomson-1 Locally

vLLM + PegaFlow: KV Cache That Survives Restarts (Hands-On)

Qwen3.8-27B Obliterated: Dont Use this Model in Production

LocateAnything: NVIDIA’s New AI Sees EVERYTHING: Run Locally

Oh Baby! Qwen3.8-27B Coming - Lets Test Qwen3.8-Max Now

Luce Spark: Run a 35B Model Under 16GB VRAM Locally

Qwen3.8-27B Ridge: Smarter Quantization, Full Power on 12GB

Nex-N2: Agentic Model with Agentic Thinking for Real-world Productivity

INTELLECT-3: A 100B+ MoE Trained with Large-Scale RL on GLM 4.5 AIR

🔇 TAP TO UNMUTE