About this tag
Model quantization is a recurring topic in WindowsForum.com discussions about running local AI models on Windows PCs. The tagged content focuses on practical sizing decisions before downloading a model, including quantization format, context-memory requirements, total and active parameter counts, and documented context limits. A thread referencing XDA Developers coverage of slowdowns and memory pressure on an RTX 4070 Ti Super highlights why tokens-per-second results matter less than whether a model fits a machine's memory budget. References to Hugging Face GGUF documentation and Qwen model documentation show how quantization choices connect to real Windows local AI workloads and hardware planning.
  1. WindowsForum AI

    Bonsai 27B on 8GB GPUs: VRAM, Context Limits, LM Studio Setup and Real-World Results

    Bonsai 27B is a 27-billion-parameter model that fits on an 8GB graphics card, but that depends on the quantization, the context length and whether you use image input. XDA-Developers writer Nolen Jonker has swapped his usual Qwen 3.5 9B and Gemma 4 E4B for it. The weight file is small. The...
  2. WindowsForum AI

    Run Local AI on an 8GB GPU: LM Studio VRAM, Context and KV Cache Tuning

    The AI model release cycle has a clear message for owners of mainstream graphics cards: bring more VRAM. Headline models often assume a 24GB card, and a lot of people are sitting on 8GB GPUs they bought for gaming, video editing or ordinary Windows work. A new hands-on piece from XDA Developers...
  3. WindowsForum AI

    Windows Local AI: Check Context Memory Before Download

    Before downloading a local AI model for a Windows PC, check its context-memory requirements, exact quantization format, total and active parameter counts, and documented context limits. A fast tokens-per-second result is useful only if the model can handle your workload within your machine’s...