Gemma 4 models already downloaded more than 335K+ times in @lmstudio 🤯

Apr 4, 2026 · 9:31 AM UTC

8
6
127
10,467
Sort replies: Relevant Recent Liked
Replying to @yagilb @lmstudio
Yes . Twice for me : the mlx version that doesn't work and the gguf that works
1
133
Try > lms runtime update mlx
1
3
79
more replies
Replying to @yagilb @lmstudio
lms load google/gemma-4-31b is causing Linux kernal panic on the same machine where lms load qwen/qwen3.5-27b worked. Why ?
1
252
try > lms runtime update Re kernel panic- not expected, we do everything in user space. Are you on a custom Linux build or something like that?
1
2
142
more replies
Replying to @yagilb @lmstudio
LM Studio hurts Gemma reputation a lot with multiple buggy implementations, today they fixed many issues, before it was full of bugs, hogged all VRAM for context, also quality was way worse
1
221
“Today they fixed many issues” is the key! We’ll continue pushing improvements
2
179
Replying to @yagilb @lmstudio
crazy
4
1
49
6,200
Replying to @yagilb @jetha @lmstudio
Everyone’s trying to find one that works
1
77
Replying to @yagilb @lmstudio
I'm using Ollama to do some benchmarks on vintage hardware. 2017 iMac · Intel Core i7 @ 4.2 GHz (4 cores, 8 threads) · 40 GB RAM · CPU-only inference Should or would LMStudio be better? I might still try it (if it can even run on intel).
96
Replying to @yagilb @lmstudio

ALT tim and eric GIF
25
Replying to @yagilb @lmstudio
when mlx support
22