LLM
GA
Open weights
Llama 4 Scout
Huge document collections up to 10M tokens
Release2025-04-05
Context10M
Max output—
Parameters109B
Local / VRAM64–80 GB
LicenseOpen weights
Overview
Llama 4 Scout is a model in the LLM category from Meta AI. It was released in 2025-04-05. The vendor is based in US. Best for: Huge document collections up to 10M tokens.
Hardware: Q4 ~60-65 GB; single H100 80 GB class.
Strengths
- 10M context
- Native multimodal
- Can run on a single server
Limits
- Multimodal license restriction for EU/UK developers
- Quality can drop at long context
Specifications
| Vendor | Meta AI |
|---|---|
| Category | LLM |
| Release | 2025-04-05 |
| Status | GA |
| License | Open weights |
| License name | Llama 4 Community License |
| Commercial use | Conditional |
| Parameters | 109B |
| Active parameters | 17B |
| Architecture | MoE transformer |
| Context window | 10M |
| Input modalities | Text, Image |
| Output modalities | Text |
| Prompt types | Text |
| Negative prompts | No |
| Max outputs | 1 |
| Local deployment | Yes |
| VRAM | 64–80 GB |
| System RAM | 96–192 GB |
| API access | Yes |
| Pricing | Free weights |
| Confidence | High |
| Vendor country | US |
| Vendor type | Frontier lab |
Pricing & hardware
Free weights
This model can be self-hosted or run locally. Approximate GPU memory: 64–80 GB. Approximate system RAM: 96–192 GB. API access is available.