Llama 4 Maverick — Specifications
| Desenvolvedor | Meta |
|---|
| Tipo | Multimodal (MoE) |
|---|
| Modalidade | Text, Image → Text |
|---|
| Parâmetros | 400B total / 17B active (MoE) |
|---|
| Janela de contexto | 1M |
|---|
| Saída máxima | — |
|---|
| Licença | Llama 4 Community (EU-restricted) |
|---|
| Pesos abertos | ✅ Yes |
|---|
| Lançamento | 2025 |
|---|
| Input price | $0.15 /1M |
|---|
| Output price | $0.6 /1M |
|---|
| API providers | Meta, Together, OpenRouter |
|---|
🖥️ Run it locally
| VRAM (4-bit) | ~240 GB |
|---|
| Minimum GPU | Multi-GPU server |
|---|
Official page →
The larger Llama 4 — 400B total / 17B active (128 experts), natively multimodal, 1M context. Server-class, but fully open under the Llama 4 Community License (EU-restricted).
