RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can Phi 3.5 vision instruct run on the RTX 4070?

Yes. At 8K context the best fit is Q8_0, using about 7.6 GB of the 11.4 GB usable on the RTX 4070. Estimated speed ceiling: ~114 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K)3.4 GB Fits4.9 GB Fits13.9 GB No49.9 GB No324
3-bit (Q3_K_M)3.8 GB Fits5.3 GB Fits14.3 GB No50.3 GB No249
4-bit (Q4_K_M)4.3 GB Fits5.8 GB Fits14.8 GB No50.8 GB No199
5-bit (Q5_K_M)4.7 GB Fits6.2 GB Fits15.2 GB No51.2 GB No171
6-bit (Q6_K)5.1 GB Fits6.6 GB Fits15.6 GB No51.6 GB No148
8-bit (Q8_0)6.1 GB Fits7.6 GB Fits16.6 GB No52.6 GB No114
16-bit (F16/BF16)9.7 GB Fits11.2 GB Tight20.2 GB No56.2 GB No61

• real GGUF file size. Speed is a bandwidth ceiling (504 GB/s), not a benchmark. Methodology.

More

Phi 3.5 vision instruct on every GPU and Mac · Every model that runs on the RTX 4070 · Other models that fit 12 GB VRAM · Try other settings in the calculator

Data: Hugging Face Hub (microsoft/Phi-3.5-vision-instruct), NVIDIA specs. Data as of 2026-11-17.