MetaLlama 4 Scout
Llama 4 Scout is a Meta open-weights model, released 5 Apr 2025. Its best configuration ranks #676 of 785 on the BenchLeader Index at 39.7 ±4.0, in the lower half. The ± is the point: 249 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in human preference (48) and lowest in coding (23). At $0.300 per million tokens blended it is cheaper than most ranked models. Output speed of 105 tokens per second puts it faster than most, with a first answer in 0.8 s. Last measured 22 Oct 2026.
- Blended price
- $0.300/M
- $0.180 in · $0.660 out
- Output speed
- 105 tok/s
- measured by Artificial Analysis
- First answer
- 0.85 s
- first token 0.64 s
- Context
- 328k
- Full answer
- 5.61 s
- median, reasoning included
- Cost per run
- $0.112
- one full Intelligence Index run
- Released
- 5 Apr 2025
- Overall index40
- Reasoning37
- Coding23
- Agents & tools36
- Maths46
- Knowledge38
- Instruction following47
- Human preference48
- Multimodal41
- Long context34
- Composite37
How its index has moved
8 Sept 2026 to 20 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. 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 | not stated | Source | Trend |
|---|---|---|---|
| GPQA Diamond | 51.8%#211 | Epoch AI Benchmarking Hub | |
| Arena Hard Prompts | 1329#233 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 58.7%#385 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 3.8%#567 | Artificial Analysis | |
| Kagi LLM Benchmark | 36.9%#114 | Kagi LLM Benchmark | |
| GPQA Diamond (HELM)not in index | 50.7%#37 | HELM Capabilities | |
| ARC-AGI-1 | 0.5%#255 | ARC Prize | |
| ARC-AGI-2 | 0.0%#242 | ARC Prize | |
| CritPt | 0.0%#315 | Artificial Analysis | |
| LMCAnot in index | 12.0%#176 | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | 57.9%#192 | Epoch AI Benchmarking Hub | |
| ForecastBenchnot in index | 59.1%#74 | ForecastBench |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| SciCode | 21.3%#212 | SciCode | |
| Arena Coding | 1363#229 | Arena (LMArena) | |
| SciCode (AA)not in index | 21.3%#226 | Artificial Analysis | |
| SWE-bench Verified (bash only) | 9.1%#44 | SWE-bench | |
| SWE-bench Verified (any scaffold)not in index | 9.1%#60 | SWE-bench | |
| Terminal-Bench 4.0 (AA)not in index | 0.0%#152 | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 3.8%#213 | Artificial Analysis |
Agents & tools
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Terminal-Bench Hard | 1.5%#338 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 15.5%#360 | Artificial Analysis | |
| BFCL Overall | 28.1%#54 | Berkeley Function Calling Leaderboard | |
| GDPval-AA v2.1not in index | 0.0%#124 | Artificial Analysis | |
| τ³-Banking (AA) | 3.3%#198 | Artificial Analysis | |
| AutomationBenchnot in index | 0.2%#219 | Artificial Analysis | |
| GDP.pdf (AA)not in index | 0.0%#220 | Artificial Analysis |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| OTIS Mock AIME | 7.8%#234 | Epoch AI Benchmarking Hub | |
| MATH Level 5 | 62.3%#46 | Epoch AI Benchmarking Hub | |
| Omni-MATH (HELM) | 37.3%#34 | HELM Capabilities | |
| Arena Maths | 1309#223 | Arena (LMArena) |
Knowledge
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-Omniscience | -52.1#406 | Artificial Analysis | |
| MMLU-Pro (HELM)not in index | 74.2%#30 | HELM Capabilities | |
| AA-Omniscience: accuracynot in index | 15.2%#418 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 20.6%#275 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| IFBench | 39.5%#267 | Artificial Analysis | |
| IFEval (HELM) | 81.8%#38 | HELM Capabilities | |
| WildBench (HELM) | 77.9%#47 | HELM Capabilities | |
| Arena Instruction Followingnot in index | 1299#239 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1322#234 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1289#228 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1319#227 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1325#232 | Arena (LMArena) |
Multimodal
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Vision | 1118#124 | Arena (LMArena) | |
| MMMU-Pro | 53.0%#234 | Artificial Analysis | |
| MMMU (validation) | 69.4%#23 | MMMU | |
| VTBnot in index | 1.6%#20 | Scale AI SEAL | |
| SpatialViz-Benchnot in index | 34.2%#4 | Epoch AI Benchmarking Hub |
Long context
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Fiction.LiveBench 120k | 27.3%#32 | Fiction.live | |
| AA-LCR | 27.7%#403 | Artificial Analysis |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 129.6#146 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 6.5#527 | Artificial Analysis | |
| HELM Capabilities meannot in index | 64.4%#35 | HELM Capabilities |
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 |
|---|---|---|---|---|---|---|
| DeepInfra | 47 tok/s | 0.59 s | $0.100 | $0.300 | 328k | fp8 |
| DeepInfra | 37 tok/s | 0.71 s | $0.150 | $0.450 | 328k | fp8 |
| DeepInfra | 31 tok/s | 1.37 s | $0.080 | $0.240 | 328k | fp8 |
Price history
The price above, per 1M tokens, every day it changed.
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 | $0.0003 | – | 3.7 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0026 | – | 6.6 s |
| Code edit | 6,000 / 1,500 | $0.0021 | – | 15.1 s |
| Agentic coding session | 60,000 / 4,000 | $0.013 | – | 38.9 s |
| Structured extraction | 2,000 / 200 | $0.0005 | – | 2.8 s |
More from Meta
See also
Questions people ask
- What are Llama 4 Scout's benchmark scores?
- As of 24 Oct 2026, Llama 4 Scout scores 39.7 ±4.0 on the BenchLeader Index, strongest in human preference (48) and weakest in coding (23). That rests on 27 independent benchmark results, and every score on this page links to its publisher.
- How does Llama 4 Scout rank?
- Llama 4 Scout is #676 of 785 ranked configurations on 24 Oct 2026. 249 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Llama 4 Scout cost per 1M tokens?
- As of 24 Oct 2026, Llama 4 Scout costs $0.180 per 1M input tokens and $0.660 per 1M output tokens, $0.300 blended (three input tokens per output token). The list price last changed on 13 Oct 2026.
- When was Llama 4 Scout released?
- Meta released Llama 4 Scout on 5 Apr 2025.
Data as of 24 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Llama 4 Scout: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/llama-4-scout.