Llama 4 Scout
Meta · Released Apr 2025
Meta's smaller open-weight MoE model (17B active / 109B total). Supports up to 10M context on some providers.
- Strengths
- Smaller open MoE supporting up to a 10M-token context on some hosts.
- Best for
- Extreme long-context tasks and budget self-hosting.
- Limitations
- Lighter than Maverick; hosted pricing and context limits vary by provider.
Input / 1M
$0.08
Output / 1M
$0.3
Cached input / 1M
—
Context window
1M
Price history
Snapshots
| Effective | Input | Output | Cached in | Note | Source |
|---|---|---|---|---|---|
| 5 Apr 2025 | $0.08 | $0.3 | — | Representative hosted rate (DeepInfra); varies by provider | pricepertoken.com |