Best local AI models for AMD HD 8790M

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 AMD Radeon HD 8790M is an older mobile graphics card equipped with 2 GB of GDDR5 video memory. This hardware limit dictates which artificial intelligence models can run directly on your GPU. To run local models successfully, the model files must be small enough to fit inside this 2 GB VRAM limit. If a model exceeds this capacity, your system must use alternative execution strategies.

The quantization column indicates the compression level applied to each model. Quantization reduces the precision of model weights to save memory. For example, a Q4_K_M quantization represents a medium four bit compression. This allows larger models like the Allegro 2.8B or Open-Sora Plan 2.7B to fit into 2 GB of VRAM. Higher quality quants like Q6_K or Q8_0 offer better accuracy but require more memory per parameter, which limits you to smaller models like the Moondream 2 1.9B or TinyLlama 1.1B.

When a model size exceeds your 2 GB VRAM, you can use CPU offloading if your computer has enough system RAM. Assuming your system has 32 GB of system RAM, you can split the workload. This approach allows you to run larger models like the SmolLM3 3B or MusicGen which need 2.2 GB or 2.4 GB of VRAM at Q4_K_M. The remaining data spills over into your system RAM, requiring 4.2 GB or 4.4 GB of system memory respectively.

CPU offloading comes with a performance cost. Transferring data between your system RAM and the AMD HD 8790M over the system bus is much slower than keeping all data inside the GDDR5 memory. While offloading allows you to run advanced models like Stable Diffusion XL or SDXL Turbo, your generation speeds will drop significantly compared to models that fit entirely within the 2 GB GPU limit.

You must also consider the context window size when running local text models. The memory figures listed are calculated using a baseline 4k context window. If you increase the context length to process longer documents or conversations, the memory requirements will rise. A model that fits perfectly at a 4k context window may exceed your 2 GB VRAM limit and trigger slow CPU offloading if you attempt to use longer context lengths.