Llama 4 Scout
Language Available Full comparison ↗ meta/llama-4-scout · by Meta
· mixture-of-experts
Pricing — 1 offering(s)
Input tokens
- $0.10 / 1M tokens (input) Historical 2025-04-05 → 2026-07-21
- $0.18 / 1M tokens (input) Current 2026-07-22 → present
Output tokens
- $0.30 / 1M tokens (output) Historical 2025-04-05 → 2026-07-21
- $0.59 / 1M tokens (output) Current 2026-07-22 → present
Showing the active price and any recorded history. Full pricing history is available via the paid API — see API docs.
See how Llama 4 Scout fits into a cost-aware routing setup
See how →Capability profile
How Llama 4 Scout rates across core capability dimensions, with the task-level evidence behind each rating.
Solid for its cost tier; below Maverick on the hardest multi-step tasks.
10M-token context — the longest available in any model at this price point.
MoE efficiency: high throughput at very low cost per token.
Coding benchmarks
View the full coding benchmark leaderboard →| Benchmark | Score | Harness | Source |
|---|---|---|---|
| swe-bench-verified ↗ | 9.1% 2025-07 | mini-swe-agent 0.0.0 | 📊 leaderboard ↗ |
| bigcodebench ↗ | 16.9% 2025-04 | — | 📊 leaderboard ↗ |
Operator guidance
Route here specifically for workloads that need a very long context window and cannot afford frontier closed-model pricing. The 10M-token context is the primary differentiator — if your task doesn't need it, Llama 4 Maverick or GPT-4o mini are better-rounded alternatives at similar cost.
Use cases
- Ultra-long-context tasks: entire codebases, legal corpora, full books
- Multimodal workflows (image + text) at minimal cost
- Open-weights hosting where context length is the primary constraint
Limitations
- Newer model; fewer production deployments than Llama 3.x series
- Capability ratings are qualitative, not a single cited benchmark run