Google's latest, with vision and 128K context. Strong quality per parameter.
chatgeneralvision · sizes: 1B, 4B, 12B, 27B · Gemma
| machine | best fit | verdict | speed |
|---|---|---|---|
| Apple M1 · 8GB | 1B Q8_0 | ✅ Runs comfortably | 35–55 tok/s |
| Apple M1 Pro · 16GB | 4B Q8_0 | ✅ Runs comfortably | 20–35 tok/s |
| Apple M2 · 16GB | 4B Q8_0 | ✅ Runs comfortably | 11–18 tok/s |
| Apple M3 Pro · 18GB | 12B Q3_K_M | ✅ Runs (tight) | 9–15 tok/s |
| Apple M2 Pro · 32GB | 27B Q3_K_M | ✅ Runs (tight) | 6–10 tok/s |
| Apple M4 Pro · 48GB | 27B Q8_0 | ✅ Runs (tight) | 4–7 tok/s |
| Apple M3 Max · 36GB | 27B Q4_K_M | ✅ Runs (tight) | 9–14 tok/s |
| Apple M2 Max · 64GB | 27B Q8_0 | ✅ Runs comfortably | 6–9 tok/s |
| Apple M4 Max · 128GB | 27B Q8_0 | ✅ Runs comfortably | 7–12 tok/s |
| Apple M2 Ultra · 192GB | 27B Q8_0 | ✅ Runs comfortably | 8–13 tok/s |
| NVIDIA GeForce RTX 3060 · 12GB VRAM | 12B Q4_K_M | ✅ Runs (tight) | 14–25 tok/s |
| NVIDIA GeForce RTX 4070 · 12GB VRAM | 12B Q4_K_M | ✅ Runs (tight) | 20–35 tok/s |
| NVIDIA GeForce RTX 4090 · 24GB VRAM | 27B Q4_K_M | ✅ Runs (tight) | 20–35 tok/s |
| NVIDIA GeForce RTX 5080 · 16GB VRAM | 12B Q5_K_M | ✅ Runs (tight) | 35–55 tok/s |
| NVIDIA GeForce RTX 5090 · 32GB VRAM | 27B Q5_K_M | ✅ Runs comfortably | 35–55 tok/s |
| AMD Radeon RX 7900 XTX · 24GB VRAM | 27B Q4_K_M | ✅ Runs (tight) | 18–30 tok/s |
Shopping for a machine? What hardware do I need to run Gemma 3? →
Check your exact rig: open the checker →
3 models share Gemma 3's KV-cache geometry — they size memory the same way, so a rig that fits one tends to fit its same-size kin.
Explore the whole catalog by architecture: open the graph →