Best local AI models for 20GB VRAM

Ranked by what actually fits at 32K context, computed from real file bytes.

A 20GB card gives you about 18.60 GiB to work with after driver overhead. 2 indexed models fit at 32K context — the largest being HunyuanImage-2.1 at 17.5B parameters in IQ3_S.

From the file· fit from summed bytesFrom the file· KV per layer

Fits in 20GB at 32K context

largest quantization that fits, per model
ModelModalityBest quantParamsTotalHeadroom
HunyuanImage-2.1image generationIQ3_S17.5B17.36 GiB1.24 GiB
Janus-Pro-7Bimage generationI1-IQ3_XXS7.4B18.40 GiB0.20 GiB
Spec sheetPredictedwhat these mean

This page models a generic 20GB accelerator, so it answers what fits rather than how fast it runs. For tokens per second you need a specific card — pick one from hardware, where bandwidth is known.