Best local AI models for NVIDIA GT 750M

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 GT 750M is a legacy mobile graphics card equipped with 2 GB of GDDR5 memory. This hardware configuration limits the size of the artificial intelligence models you can run locally. To fit inside the tight 2 GB video memory boundary, models must use quantization. Quantization is a compression method that reduces the precision of model weights to save space. The quant column shows the best balance of size and quality for each model.

For models that fit entirely within the 2 GB limit, you can run execution directly on the graphics card. The 2.8B Allegro model fits at the Q4_K_M quantization level and uses exactly 2 GB of video memory. The 2.7B Open-Sora Plan also fits at Q4_K_M and uses 2 GB. Smaller models like the 2.6B LFM2 and 2.6B Playground v2.5 use 1.9 GB of video memory at the Q4_K_M quantization level. Stable Diffusion 3.5 Medium and Canary 1B / Qwen-2.5B both require 1.8 GB of video memory at Q4_K_M.

Slightly smaller models can use higher precision quantization levels for better output quality. SeamlessM4T v2, Parler-TTS, and Kimi K3 DSpark run at the Q5_K_M quantization level and use between 1.9 GB and 2 GB of video memory. Models like SmolVLM 256M / 500M / 2B, Stable Diffusion 3 Medium, Pyramid Flow, and Wav2Vec2 / XLS-R can run at the Q6_K quantization level while using exactly 2 GB of video memory. Moondream 2 uses 1.9 GB at Q6_K, while Qwen3 1.7B and SmolLM2 135M / 360M / 1.7B use 1.7 GB at Q6_K.

Very small models can run at the high quality Q8_0 quantization level. StableLM 2 1.6B, Sana 0.6B / 1.6B, Zonos 0.1, Dia 1.6B, and Whisper Large v3 all use exactly 2 GB of video memory at Q8_0. ControlNet / T2I-Adapter / IP-Adapter, Hunyuan-DiT, Stable Video Diffusion, Whisper Large v2 / turbo, AudioGen, and AudioLDM 2 use 1.9 GB at Q8_0. Tango 2 uses 1.8 GB at Q8_0. TinyLlama 1.1B and SantaCoder 1.1B run comfortably at Q8_0 while using only 1.4 GB of video memory.

When a model is too large for the 2 GB video memory, you must use CPU offload. This technique splits the model between the graphics card and your system RAM. CPU offload requires a system with at least 32 GB of system RAM. Offloading makes processing much slower because data must travel between the system RAM and the graphics card. Models like SmolLM3 3B, Replit Code v1.5 3B, Kandinsky 3.1, Voxtral Mini / Small, Orpheus TTS, and Higgs Audio v2 need 2.2 GB of video memory at Q4_K_M and require 4.2 GB of system RAM.

Larger offloaded models require even more resources. MusicGen small/medium/large needs 2.4 GB of video memory at Q4_K_M and 4.4 GB of system RAM. Stable Diffusion XL needs 4.1 GB of video memory at FP8 / optimized and 6.1 GB of system RAM. SDXL Turbo and SDXL Lightning both need 2.6 GB of video memory at Q4_K_M and 4.6 GB of system RAM. You must also remember the 4k context caveat. Running text models with long context windows increases memory usage quickly and can cause out of memory errors on this card.