MetaLlama 4 Scout
Best configuration ranks #521 of 610 on the BenchLeader Index at 39.7 ±4.7. Last measured 2 Sept 2026. Released 5 Apr 2025.
- Blended price
- $0.283/M
- $0.180 in · $0.590 out
- Output speed
- 84 tok/s
- First answer
- 0.86 s
- first token 0.91 s
- Context
- 328k
- Overall index40
- Reasoning36
- Coding23
- Agents & tools39
- Maths40
- Knowledge39
- Instruction following42
- Human preference49
- Multimodal43
- Long context35
- Composite37
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source |
|---|---|---|
| GPQA Diamond | 51.8%#196 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1328#210 | LMArena |
| GPQA Diamond (AA)not in index | 58.7%#366 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 3.8%#482 | Artificial Analysis |
| Kagi LLM Benchmark | 36.9%#114 | Kagi LLM Benchmark |
| ARC-AGI-1 | 0.5%#185 | ARC Prize |
| ARC-AGI-2 | 0.0%#174 | ARC Prize |
Coding
| Benchmark | default | Source |
|---|---|---|
| SciCode | 17.0%#155 | SciCode |
| LMArena Coding | 1362#208 | LMArena |
| SciCode (AA)not in index | 21.3%#158 | Artificial Analysis |
| SWE-bench Verified (bash only) | 9.1%#43 | SWE-bench |
| SWE-bench Verified (any scaffold)not in index | 9.1%#64 | SWE-bench |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| Terminal-Bench Hard | 1.5%#321 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 15.5%#345 | Artificial Analysis |
| BFCL Overall | 28.1%#54 | Berkeley Function Calling Leaderboard |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 7.8%#215 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 62.3%#45 | Epoch AI Benchmarking Hub |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -52.1#333 | Artificial Analysis |
Instruction following
| Benchmark | default | Source |
|---|---|---|
| IFBench | 39.5%#257 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1321#212 | LMArena |
Multimodal
| Benchmark | default | Source |
|---|---|---|
| LMArena Vision | 1118#105 | LMArena |
| MMMU-Pro | 53.0%#199 | Artificial Analysis |
| MMMU (validation) | 69.4%#22 | MMMU |
Long context
| Benchmark | default | Source |
|---|---|---|
| Fiction.LiveBench 120k | 27.3%#32 | Fiction.live |
| AA-LCR | 27.7%#331 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| Epoch Capabilities Indexnot in index | 129.6#133 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 6.5#443 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Google Vertex | 40 tok/s | 0.53 s | $0.250 | $0.700 | 1.3M | – |
| DeepInfra | 15 tok/s | 0.91 s | $0.100 | $0.300 | 328k | fp8 |
| NovitaAI | 14 tok/s | 1.52 s | $0.180 | $0.590 | 131k | bf16 |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0002 | – | 4.4 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0025 | – | 8.0 s |
| Code edit | 6,000 / 1,500 | $0.0020 | – | 18.7 s |
| Agentic coding session | 60,000 / 4,000 | $0.013 | – | 48.4 s |
| Structured extraction | 2,000 / 200 | $0.0005 | – | 3.2 s |
See also
Data as of 9 Sept 2026. Compare with another model.