Best local AI models for AMD R9 M275

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 R9 M275 is a mobile graphics card equipped with 2 GB of GDDR5 memory. This physical memory limit determines which artificial intelligence models can run directly on your hardware. When running local models, the entire active weight set must fit within this space to maintain reasonable processing speeds.

To fit larger models into the 2 GB limit, you must use quantized files. The quant column indicates the compression level applied to the model weights. For example, Q4_K_M represents a four bit quantization that reduces file size while preserving most of the original model accuracy. Higher quants like Q6_K and Q8_0 offer better precision but require more memory space.

Several models can run entirely within your hardware limits. The Allegro 2.8B model fits at Q4_K_M quantization using exactly 2 GB of memory. The Open-Sora Plan 2.7B model also uses 2 GB at Q4_K_M. For slightly smaller footprints, the LFM2 2.6B and Playground v2.5 2.6B models require 1.9 GB of memory at Q4_K_M. Stable Diffusion 3.5 Medium 2.5B and Canary 1B / Qwen-2.5B use 1.8 GB at the same quantization level.

Other architectures utilize different quantization levels to balance performance. SeamlessM4T v2 2.3B and Kimi K3 DSpark 2.2B use 2 GB of memory at Q5_K_M. Parler-TTS 2.2B fits within 1.9 GB at Q5_K_M. Models like SmolVLM 2B, Stable Diffusion 3 Medium 2B, Pyramid Flow 2B, and Wav2Vec2 / XLS-R 2B can run at Q6_K using 2 GB. Moondream 2 1.9B fits at Q6_K using 1.9 GB, while Qwen3 1.7B and SmolLM2 1.7B use 1.7 GB at Q6_K.

Highly precise Q8_0 models are also compatible. StableLM 2 1.6B, Sana 1.6B, Zonos 0.1, Dia 1.6B, and Whisper Large v3 1.55B all run at Q8_0 using 2 GB. ControlNet / T2I-Adapter / IP-Adapter 1.5B, Hunyuan-DiT 1.5B, Stable Video Diffusion 1.5B, Whisper Large v2 / turbo 1.5B, AudioGen 1.5B, and AudioLDM 2 1.5B use 1.9 GB at Q8_0. Tango 2 1.4B uses 1.8 GB at Q8_0, while TinyLlama 1.1B and SantaCoder 1.1B use 1.4 GB at Q8_0.

When a model exceeds the 2 GB memory limit, you must offload data to your system RAM. Assuming you have 32 GB of system RAM, you can run larger models with a speed penalty. SmolLM3 3B, Replit Code v1.5 3B, Kandinsky 3.1, Voxtral Mini / Small, Orpheus TTS, and Higgs Audio v2 3B require 2.2 GB of video memory at Q4_K_M and 4.2 GB of system RAM. MusicGen 3.3B needs 2.4 GB of video memory and 4.4 GB of system RAM. Stable Diffusion XL 3.417B needs 4.1 GB at FP8 and 6.1 GB of system RAM. SDXL Turbo 3.5B and SDXL Lightning 3.5B require 2.6 GB of video memory and 4.6 GB of system RAM.

Be aware of the context window limits when running text models. The memory figures listed assume a standard 4k context window. Increasing the context length requires additional memory for the key value cache. This extra demand can quickly exceed your 2 GB limit and force slow system RAM offloading.