Gemma 3 4B — Specifications
| Desenvolvedor | |
|---|---|
| Tipo | LLM (multimodal) |
| Modalidade | Text, Image → Text |
| Parâmetros | 4B |
| Janela de contexto | 128K |
| Saída máxima | — |
| Licença | Gemma (open) |
| Pesos abertos | ✅ Yes |
| Lançamento | 2025 |
| Input price | $0.05 /1M |
| Output price | $0.1 /1M |
| API providers | Google AI Studio, Ollama |
🖥️ Run it locally
| VRAM (4-bit) | ~3 GB |
|---|---|
| Minimum GPU | Any 6GB+ GPU |
A compact multimodal Gemma 3 (text + image, 128K) that runs on almost any modern GPU — ~3GB at 4-bit. Excellent for low-VRAM and edge setups.