Models › d3-flash
d3-flash: VRAM needs and speed by GPU
At Q4_K_M, d3-flash needs about 7.6 GB of VRAM. No measured runs yet. Speeds below are estimates. Submit a run.
8.95B 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 |
|---|---|---|---|
| Q4_K_M | 6.6 GB | 7.6 GB | 12 GB cards |
| Q8_0 | 9.5 GB | 10.5 GB | 12 GB cards |
| BF16 | 17.9 GB | 18.9 GB | 24 GB cards |
Expected speed on popular hardware
| Hardware | Best quant that fits | Expected tok/s |
|---|---|---|
| NVIDIA RTX 3060 | Q8_0 | 32.3 estimate |
| NVIDIA RTX 4060 Ti | Does not fit | |
| NVIDIA RTX 3090 | Q8_0 | 68.9 estimate |
| NVIDIA RTX 4090 | Q8_0 | 76.3 estimate |
| NVIDIA RTX 5090 | Q8_0 | 121 estimate |
| AMD Radeon RX 7900 XTX | Q8_0 | 76.6 estimate |
| Apple M4 Pro (20-core GPU) (64 GB) | Q8_0 | 24.3 estimate |
| Apple M3 Max (40-core GPU) (64 GB) | Q8_0 | 35.5 estimate |
| AMD Ryzen AI Max+ 395 (Radeon 8060S) | Q8_0 | 21.8 estimate |
Too big for your card?
Run d3-flash 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.