Blog hub
Practical guides around what users actually search for: what fits in your VRAM, how to choose models on Mac, how quantization works, which coding and vision models to run locally.
Video Memory Guide
From 6GB, 8GB, 12GB, 24GB to 48GB, explain how model parameter volume, quantized version, KV cache and system overhead combine to determine whether it can be loaded.
Read articleApple chip
Explain why the total memory on Mac cannot be used as video memory, and how to choose the suitable model for 16GB, 32GB, 64GB, and 128GB machines.
Read articleQuantify
The more common GGUF quantification of memory usage, quality loss and speed trade-off helps users understand the three preferences of quality priority, balance and long context.
Read articleprogramming model
From the four scenarios of code generation, interpretation, reconstruction and long context, explain why programming purposes cannot just look at model size and download volume.
Read articlemultimodal
This article introduces the additional issues of graphics memory, image encoder, context, and reasoning back-end support that need to be considered in visual models compared to text models.
Read articleTool selection
Explain to ordinary users the installation experience, model management, performance tuning and applicable groups of three common local running methods.
Read article