3 ms·
I switched to GLM-5.3 flash on high for this reason. Too many "but wait" in the Deepseek-v4 reasoning.
by iamniels 24d ago
I switched to GLM-5.3 flash on high for this reason. Too many "but wait" in the Deepseek-v4 reasoning.
- alfiedotwtf 24d agoGLM 5.3 Flash was the killer for me. It really feels like we have Claude-approaching models at home. I just wish they kept parameter count down in order to fit entirely within commonly used RAM sizes