Gemma 3 12B — Specifications
| Développeur | Google |
|---|
| Type | LLM (multimodal) |
|---|
| Modalité | Text, Image → Text |
|---|
| Paramètres | 12B |
|---|
| Fenêtre de contexte | 128K |
|---|
| Sortie maximale | — |
|---|
| Licence | Gemma (open) |
|---|
| Poids ouverts | ✅ Yes |
|---|
| Date de sortie | 2025 |
|---|
| Input price | $0.05 /1M |
|---|
| Output price | $0.15 /1M |
|---|
| API providers | Google AI Studio, OpenRouter, Ollama |
|---|
🖥️ Run it locally
| VRAM (4-bit) | environ 8 Go |
|---|
| Minimum GPU | RTX 4070 12GB |
|---|
Official page →
A mid-size multimodal Gemma 3 (text + image, 128K, 140+ languages) that runs 4-bit on a 12GB GPU — a great balance of capability and footprint.
