Can Phi 3.5 vision instruct run on the RTX 3070?
Yes. At 8K context the best fit is Q6_K, using about 6.6 GB of the 7.6 GB usable on the RTX 3070. Estimated speed ceiling: ~132 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | 32K ctx | 128K ctx | Max tok/s (est., 8K) |
|---|---|---|---|---|---|
| 2-bit (Q2_K) | 3.4 GB Fits | 4.9 GB Fits | 13.9 GB No | 49.9 GB No | 288 |
| 3-bit (Q3_K_M) | 3.8 GB Fits | 5.3 GB Fits | 14.3 GB No | 50.3 GB No | 221 |
| 4-bit (Q4_K_M) | 4.3 GB Fits | 5.8 GB Fits | 14.8 GB No | 50.8 GB No | 177 |
| 5-bit (Q5_K_M) | 4.7 GB Fits | 6.2 GB Fits | 15.2 GB No | 51.2 GB No | 152 |
| 6-bit (Q6_K) | 5.1 GB Fits | 6.6 GB Fits | 15.6 GB No | 51.6 GB No | 132 |
| 8-bit (Q8_0) | 6.1 GB Fits | 7.6 GB Tight | 16.6 GB No | 52.6 GB No | 102 |
| 16-bit (F16/BF16) | 9.7 GB No | 11.2 GB No | 20.2 GB No | 56.2 GB No | — |
• real GGUF file size. Speed is a bandwidth ceiling (448 GB/s), not a benchmark. Methodology.
More
Phi 3.5 vision instruct on every GPU and Mac · Every model that runs on the RTX 3070 · Other models that fit 8 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.