Llama 4 Scout
Open SourceOpen SourceMultimodalAPI Available
Benchmark Scores
85.9/100MMLU87.5
HumanEval89.2
Math79
Reasoning85.8
Coding88
About Llama 4 Scout
Llama 4 Scout is Meta's mixture-of-experts model with 17B active parameters and an unprecedented 10M token context window. It uses 16 experts and delivers competitive performance while being highly efficient.
Strengths
- +10M token context window
- +MoE efficiency
- +Open source
- +Strong multilingual
Weaknesses
- -New model with limited ecosystem
- -MoE can have quality variance
- -Fewer fine-tuned variants
Pricing
Per 1M tokens
Input$0.15
Output$0.15
Quick Facts
- Context Window
- 10M
- Parameters
- 17B active (109B total)
- Release Date
- 2025-04-05
- Category
- Open Source
Compare Models
See how Llama 4 Scout stacks up against other models