Home/Models/Llama 4 Scout

Llama 4 Scout

Open Source
by Meta· Released 2025-04
Official Docs ↗

Llama 4 Scout is Meta's efficient open-weight model designed for single-H100 deployment. Its 10M token context window is the largest of any commonly hosted model, making it ideal for massive document ingestion, codebase analysis, and long-context RAG.

Input Price
$0.08/1M
per 1M tokens
Output Price
$0.30/1M
per 1M tokens
Context Window
10M tokens
max tokens
ELO Score
1280
LMSYS Arena

Pricing Breakdown

VolumeInput CostOutput CostCombined (50/50)
1,000 tokens$0.0001$0.0003$0.0002
10,000 tokens$0.0008$0.0030$0.0019
100,000 tokens$0.0080$0.0300$0.0190
1,000,000 tokens$0.0800$0.3000$0.1900

Strengths

  • Runs on a single H100 — cheapest self-host target in the Llama 4 family
  • 10M token context window — industry-leading for long context
  • Open weights
  • Natively multimodal

Weaknesses

  • Weaker than Maverick on complex reasoning
  • Smaller community than Llama 3 ecosystem

Best For

long contextlow costsummarizationdata extractiondocument analysis

Compare Llama 4 Scout With

Llama 4 Scout vs GPT-5.4
OpenAI
Llama 4 Scout vs Claude Opus 4.7
Anthropic
Llama 4 Scout vs Gemini 3.1 Pro
Google
Llama 4 Scout vs o3
OpenAI
Llama 4 Scout vs Claude Sonnet 4.6
Anthropic
Llama 4 Scout vs Gemini 2.5 Pro
Google

Frequently Asked Questions

How much does Llama 4 Scout cost?

Llama 4 Scout costs $0.08/1M per 1M input tokens and $0.30/1M per 1M output tokens.

What is Llama 4 Scout's context window?

Llama 4 Scout has a context window of 10M tokens, which means it can process up to 10,000,000 tokens in a single request.

Is Llama 4 Scout open source?

Yes, Llama 4 Scout is open source. The model weights are publicly available and can be self-hosted.

What is Llama 4 Scout best used for?

Llama 4 Scout is best suited for: long-context, low-cost, summarization, data-extraction, document-analysis. Runs on a single H100 — cheapest self-host target in the Llama 4 family.

What is Llama 4 Scout's ELO score?

Llama 4 Scout has an ELO score of 1280 on the LMSYS Chatbot Arena leaderboard, placing it in the mid-tier range.