Qwen2.5 1.5B: VRAM needs and speed by GPU
At Q4_K_M, Qwen2.5 1.5B needs about 2.1 GB of VRAM. We have 1 measured speed results for it.
Measured speeds
| Chip | Generation tok/s | Prompt tok/s | Quant | Backend | Test | Source |
|---|---|---|---|---|---|---|
| Intel Arc Pro B70 | 216 | 8,048 | Q4_K_M | SYCL FA on | tg128 | PMZFX intel-arc-pro-b70-benchmarks (MIT) details |
Memory needed per quantization
| Quant | File size | Needs (estimate) | Smallest card it fits |
|---|---|---|---|
| Q4_K_M | 1.1 GB | 2.1 GB | 8 GB cards |
Expected speed on popular hardware
| Hardware | Best quant that fits | Expected tok/s |
|---|---|---|
| NVIDIA RTX 3060 | Q4_K_M | 257 estimate |
| NVIDIA RTX 4060 Ti | Q4_K_M | 204 estimate |
| NVIDIA RTX 3090 | Q4_K_M | 547 estimate |
| NVIDIA RTX 4090 | Q4_K_M | 726 estimate |
| NVIDIA RTX 5090 | Q4_K_M | 957 estimate |
| AMD Radeon RX 7900 XTX | Q4_K_M | 608 estimate |
| Apple M4 Pro (20-core GPU) (64 GB) | Q4_K_M | 193 estimate |
| Apple M3 Max (40-core GPU) (64 GB) | Q4_K_M | 282 estimate |
| AMD Ryzen AI Max+ 395 (Radeon 8060S) | Q4_K_M | 173 estimate |
Too big for your card?
Run Qwen2.5 1.5B on a rented 24-80 GB GPU by the hour.
Rent a bigger GPU on RunPodAffiliate link: if you sign up, RigBench may earn a referral credit. It never affects rankings or numbers.