What can your GPU run?
Set your GPU memory, system RAM and context length. The list sorts every synced model into Full GPU fit, Offload required or Out of memory, using the same formula as the VRAM calculator: (weights + KV cache) × 1.10.
Run more than one model at once
Pick up to three models. Their memory is added together at the context length above, then compared with your VRAM and RAM.
How these numbers are made
Model file sizes come from the synced GGUF files. The KV cache is computed from each model's attention layout at the context length you pick, then 10% is added for runtime overhead. Everything is an estimate, not a measurement. See the methodology for the full formula, or open the VRAM calculator to check one model, an Apple Silicon Mac or several GPUs.