Models › gemma-4-E2B
gemma-4-E2B: VRAM needs and speed by GPU
No measured runs yet. Speeds below are estimates. Submit a run.
4.65B parameters · license: apache-2.0 · 131,072-token context · GGUF published Jul 16, 2026 · GGUF files on Hugging Face
Memory needed per quantization
| Quant | File size | Needs (estimate) | Smallest card it fits |
|---|---|---|---|
| Q8_0 | 5.0 GB | 5.9 GB | 8 GB cards |
| BF16 | 9.3 GB | 10.2 GB | 12 GB cards |
Expected speed on popular hardware
| Hardware | Best quant that fits | Expected tok/s |
|---|---|---|
| NVIDIA RTX 3060 | Q8_0 | 62.0 estimate |
| NVIDIA RTX 4060 Ti | Q8_0 | 49.3 estimate |
| NVIDIA RTX 3090 | Q8_0 | 132 estimate |
| NVIDIA RTX 4090 | Q8_0 | 146 estimate |
| NVIDIA RTX 5090 | Q8_0 | 231 estimate |
| AMD Radeon RX 7900 XTX | Q8_0 | 147 estimate |
| Apple M4 Pro (20-core GPU) (64 GB) | Q8_0 | 46.6 estimate |
| Apple M3 Max (40-core GPU) (64 GB) | Q8_0 | 68.1 estimate |
| AMD Ryzen AI Max+ 395 (Radeon 8060S) | Q8_0 | 41.9 estimate |
Too big for your card?
Run gemma-4-E2B 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.