Best local AI models for NVIDIA MX150

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 MX150 is an entry level laptop graphics card equipped with 2 GB of GDDR5 memory. This dedicated memory size is the most critical factor when running local artificial intelligence models. To run a model entirely on your graphics hardware, the model files must fit within this 2 GB limit. If a model exceeds this limit, your system must use alternative execution methods.

The quantization column indicates the specific compression level used to shrink these models. Quantization reduces the precision of model weights to save space. For example, a Q4_K_M quant uses a four bit format to fit larger models like Allegro 2.8B or Open-Sora Plan 2.7B into 2 GB of memory. Higher quants like Q8_0 provide better quality but require more memory, limiting you to smaller models like TinyLlama 1.1B or SantaCoder 1.1B.

When a model is too large for the 2 GB video memory, you must use CPU offload. This technique splits the workload between your graphics card and your system RAM. We assume your laptop has 32 GB of system RAM for these scenarios. Offloading allows you to run larger models like SmolLM3 3B or Replit Code v1.5 3B, which need 2.2 GB of video memory at Q4_K_M and 4.2 GB of system RAM.

CPU offload comes with a significant performance cost. Sending data back and forth between the graphics card and system RAM slows down processing speeds. While you can run models like MusicGen or SDXL Turbo using this method, the generation times will be much longer than running smaller models entirely on the graphics card. Keeping models fully inside the 2 GB video memory ensures the fastest possible generation speeds.

You must also consider the 4k context window caveat when running local text models. As your conversation history grows, the memory required to track the context increases. A model that fits perfectly at startup can quickly run out of memory as you approach a 4k context limit. You should choose slightly smaller models like SmolLM2 1.7B or Qwen3 1.7B to leave enough memory headroom for longer conversations.