Models › d3-nano
d3-nano: VRAM needs and speed by GPU
No measured runs yet. Speeds below are estimates. Submit a run.
2.39B parameters · license: apache-2.0 · 262,144-token context · GGUF published Oct 12, 2026 · GGUF files on Hugging Face
Memory needed per quantization
| Quant | File size | Needs (estimate) | Smallest card it fits |
|---|---|---|---|
| Q8_0 | 2.6 GB | 3.4 GB | 8 GB cards |
| BF16 | 4.8 GB | 5.7 GB | 8 GB cards |
Expected speed on popular hardware
| Hardware | Best quant that fits | Expected tok/s |
|---|---|---|
| NVIDIA RTX 3060 | Q8_0 | 121 estimate |
| NVIDIA RTX 4060 Ti | Q8_0 | 95.9 estimate |
| NVIDIA RTX 3090 | Q8_0 | 257 estimate |
| NVIDIA RTX 4090 | Q8_0 | 285 estimate |
| NVIDIA RTX 5090 | Q8_0 | 450 estimate |
| AMD Radeon RX 7900 XTX | Q8_0 | 286 estimate |
| Apple M4 Pro (20-core GPU) (64 GB) | Q8_0 | 90.7 estimate |
| Apple M3 Max (40-core GPU) (64 GB) | Q8_0 | 132 estimate |
| AMD Ryzen AI Max+ 395 (Radeon 8060S) | Q8_0 | 81.5 estimate |
Too big for your card?
Run d3-nano 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.