Llama 4 Scout
Meta · 1 configuration
Intelligence8.1
Input / 1M$0.19
Output / 1M$0.68
Context10M
Tariff: Meta · Source: artificial-analysis · USD per million tokens
Thinking configuration
What does more thinking buy you?
Compare this model’s measured thinking configurations.
Scroll the chart ↔ · tap a point for details
Efficient frontier: best measured score at each cost1 measured configurations · 0 without both values
Selected: thinking default · Click a point to switch configuration
Llama 4 Scout 17B Instruct (16E) is a mixture-of-experts (MoE) language model developed by Meta, activating 17 billion parameters out of a total of 109B. It supports native multimodal input...
Intelligence IndexArtificial Analysis composite index, not a percentage. Different versions of the index are not directly comparable.
8.1 pointsGPQA DiamondGraduate-level scientific questions. Measures specialist reasoning rather than web-search quality.
58.7 %Humanity’s Last ExamDifficult questions across many disciplines, compared under the same evaluation setup.
3.8 %SciCodeCode generation for scientific research problems.
21.3 %Terminal-Bench 2.1Agent tasks in a terminal environment. Results also depend on the evaluation harness.
3.7 %τ²-BenchTool use and interaction in assistance scenarios. Does not replace an evaluation of your integration.
15.5 %IFBenchVerifiable instruction following. A proxy for precision rather than a benchmark of creative style.
39.5 %MMMU-ProMultimodal understanding of academic problems that include images.
52.9 %Omniscience accuracyFactual accuracy in Artificial Analysis's Omniscience evaluation.
15.2 %Specifications from the catalog
Inputtext, image
Outputtext
Tool callingYes
Structured outputYes
Speed132.7 tok/s
First token0.85 s
Maximum output16.4K
Catalog listing date2025-04-05