Can SmolVLM Instruct run on the RTX 3080 10GB?
Yes. At 8K context the best fit is F16, using about 5.4 GB of the 9.5 GB usable on the RTX 3080 10GB. Estimated speed ceiling: ~210 tokens/sec.
Memory by quant and context
| Quant | 4K ctx | 8K ctx | Max tok/s (est., 8K) |
|---|---|---|---|
| 2-bit (Q2_K) | 2.0 GB Fits | 2.8 GB Fits | 902 |
| 3-bit (Q3_K_M) | 2.3 GB Fits | 3.0 GB Fits | 692 |
| 4-bit (Q4_K_M) • | 2.3 GB Fits | 3.0 GB Fits | 683 |
| 5-bit (Q5_K_M) | 2.7 GB Fits | 3.5 GB Fits | 475 |
| 6-bit (Q6_K) | 3.0 GB Fits | 3.7 GB Fits | 413 |
| 8-bit (Q8_0) • | 3.0 GB Fits | 3.8 GB Fits | 394 |
| 16-bit (F16/BF16) • | 4.6 GB Fits | 5.4 GB Fits | 210 |
• real GGUF file size from ggml-org/SmolVLM-Instruct-GGUF. Speed is a bandwidth ceiling (760 GB/s), not a benchmark. Methodology.
More
SmolVLM Instruct on every GPU and Mac · Every model that runs on the RTX 3080 10GB · Other models that fit 12 GB VRAM · Try other settings in the calculator
Data: Hugging Face Hub (HuggingFaceTB/SmolVLM-Instruct), NVIDIA specs. Data as of 2026-11-17.