| Llama 4 Scout | Llama 4 Maverick | |
|---|---|---|
| arenaElo | 1278 | 1322 |
| humanEval | 76.3 | 82 |
| sweBench | 35 | 45 |
| mmlu | 80.4 | 85.1 |
| gpqa | 49.1 | 54.3 |
| aime | 50 | 60 |
| Price in/out ($/M) | $0.18 / $0.6 | $0.35 / $1.4 |
Scout is smaller, cheaper and offers a 10M-token context for retrieval. Maverick is bigger, smarter and the default for reasoning and agentic use. Use Scout for mass retrieval and summarization; use Maverick for reasoning-heavy tasks.