Llama 4 Scout
Open weights
Meta's small Llama 4 variant - 17B-active / 109B-total MoE that fits on a single H100, with a 10M-token context window.
- Publisher
- Meta FAIR (Fundamental AI Research)
- Family
- Llama 4
- Params
- 109B total / 17B active (16 experts)
- Context
- 1.3M
- Input
- Output
- License
- Llama Community
- Released
- Apr 2025
- Canonical
- llama.com/models/llama-4
Cite
Notes
Only stored in your browser.
Context
1.3M
Input
$0.18
/1M tok
Output
$0.66
/1M tok
Blended
$0.30
#261/ 461/1M · 3:1
Speed
133
#40/ 461tok/s
Cache read
$0.06
0.31x input
Price & speed via Artificial Analysis · ranked across tracked models · caching via OpenRouter
Reported on 13 evals across 8 domains - best 84.4% on MATH-500 · #77 of 190