So the local models should have reached that level by now?
Yes, depends on what you mean by local though, glm-5.3-flash and qwem3.8-flash-next are definitely tjere, qwen3.8;27b is arguably there too. Bit these are still heavy models and require expensive gpus to ru
1