Best local AI models for NVIDIA 930M

The largest models that fit fully

The 30 largest of the 56 models that fit; every smaller model in the catalog fits too. Best quant means the highest quality compression whose weights and 4k context both sit inside the memory.

Close, but only with CPU offload

These need more than the card holds at their smallest practical quant, so part of the model runs from system memory (figures assume 32 GB of it). They work, several times slower.

How to read this

The NVIDIA 930M is an entry level laptop graphics card equipped with 2 GB of DDR3 memory. Because of this limited memory capacity, running artificial intelligence models locally requires careful selection of model sizes and quantization levels. The memory size of 2 GB represents the absolute physical limit of what can be loaded directly onto the graphics card for fast processing. If a model exceeds this limit, it cannot run entirely on the hardware without alternative strategies.

To fit models onto this hardware, developers use quantization. The quant column indicates the specific compression format used to reduce the size of the model weights. For example, a Q4_K_M quantization compresses the model parameters to roughly four bits. This allows larger models like Allegro 2.8B or Open-Sora Plan 2.7B to fit within the 2 GB limit by using exactly 2 GB of memory. Higher quality quants like Q8_0 provide better accuracy but require more memory per parameter, limiting their use to smaller models like TinyLlama 1.1B.

When a model is too large for the 2 GB DDR3 memory, you must use CPU offload. This technique splits the model between the graphics card and your system RAM. For instance, running SmolLM3 3B or Replit Code v1.5 3B at Q4_K_M requires 2.2 GB of video memory and 4.2 GB of system RAM. Offloading allows you to run larger architectures, but it comes with a significant cost. DDR3 video memory and system RAM are much slower than modern graphics memory, which will result in much slower processing speeds.

System RAM requirements increase with the size of the model when offloading. Running Stable Diffusion XL requires 4.1 GB of video memory and 6.1 GB of system RAM using FP8 or optimized settings. Similarly, SDXL Turbo and SDXL Lightning require 2.6 GB of video memory and 4.6 GB of system RAM at Q4_K_M. You must ensure your computer has enough system RAM, such as the assumed 32 GB system RAM, to handle these split workloads without crashing.

There is an important caveat regarding the context window size when running local text models. The memory figures listed for models like SmolLM2 1.7B or Qwen3 1.7B are calculated at a standard 4k context window. If you increase the context window to process longer documents, the memory usage will grow rapidly. This extra memory demand can easily exceed the 2 GB limit of the NVIDIA 930M, forcing the system to slow down or fail to generate responses.

By selecting the correct quantization level, you can run a variety of audio, image, and text models on this hardware. Models like Moondream 2 at Q6_K use 1.9 GB of memory, while Whisper Large v3 at Q8_0 uses exactly 2 GB. Understanding the balance between model size, quantization, and offloading allows you to get the best possible performance from the NVIDIA 930M.