Post

Seems that 16GB VRAM isn't enough for vibe coding just yet. Models trained for tool calls seem to struggle with them when quantized