7h ago
••
Latest News
Latest news from Subscribe to LocalLLaMA's sources
8h ago
Qwen3.6 is confidently wrong about WASM
8h ago
Can we stop dunking on DiffusionGemma and hack it instead?
9h ago
Want to build a custom model
12h ago
Strix Halo desktop trying to compete against DGX Spark
14h ago
Codebase getting larger - Qwen3.6-27B starting to compound issues - how to work smartly with this model?
15h ago
Storing an index to a scale instead of the scale itself with Q4_0 quant reduces scale size by ~31% (small gain but interesting)
15h ago
I need a model that gets stuck in loops.
15h ago
I am losing my mind with FOMO and need some sanity checking about model capabilities
15h ago
A single federal order switched off the best cloud model overnight. Clearest case for running local I've seen yet.
15h ago
Open source AI Must Win
16h ago
Did decentralized training ever go anywhere?
17h ago
Me after installing my second P40 into my goblin box and having 48gb VRAM for the first time in my life
18h ago
Yay got Gemma 12B QAT working on old 1080ti (maybe with speculative decoding?)
18h ago
WIP EAGLE3 for Qwens
18h ago
Snapcompact: Saving Tokens With Images
19h ago
Best models in 3x3090 (72GB VRAM) in Q2 2026?
19h ago
Is there a decentralized place where model weights are stored/uploaded?
19h ago
Not looking good for GLM 5.2 Air... but maybe a flash model?
20h ago