What fits my card?

Pick your GPU or Mac. We list the open models that fit in its memory, the best quantization that fits, and the speed to expect.

My hardware Memory

Not listed? Enter it by hand

VRAM / unified memory (GB) Memory bandwidth (GB/s, optional) Type

Also show models that don't fit Sort

How to read this: "Measured" speeds link to a real run on that exact chip. Speeds marked estimate scale a real measured run on your chip by model size (generation speed is limited by memory bandwidth), or use the spec-sheet bandwidth when we have no run yet. Fit assumes the whole model on the GPU with an ~8K-token context. Details.

Every Monday: new open models grouped by how much VRAM they need, plus the newest measured benchmarks. Free, one email a week.