Llama 4 Scout

by Meta · meta-llama/llama-4-scout

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...

tool-usevisionreleased 2025-04-05

JSON Route to this model

Legal score · raw

not yet benchmarked

Context

1.31M

max output 16K

Input

$0.10

per 1M tokens

Output

$0.30

per 1M tokens

Suite cost

run the suite to see

Benchmark results

Task	Category	Raw	Juiced	Correct	Latency	Cost	Ran
Hearsay Identification	Evidence	-	-	-	-	-	-
Bluebook Citation Format	Research & Writing	-	-	-	-	-	-
Federal Civil Procedure	Procedure	-	-	-	-	-	-
Limitations Arithmetic	Procedure	-	-	-	-	-	-
Contract Clause Classification	Contracts	-	-	-	-	-	-
Citation Hallucination Resistance	Reliability	-	-	-	-	-	-

Measured by DocketBuster

These numbers come from DocketBuster's own legal battery, not from DocketRouter's suite. Latest run per metric, with n and a 95% Wilson interval where the source reports one. See docketbuster.com/benchmarks.

Metric	Value	n	Interval	Measured
Statute pinpoint, exact section (no retrieval)	9.1% (28/307)	307	95% CI 6.4% to 12.9%	2026-08-24
Statute pinpoint, exact section (with DocketBuster retrieval)	76.9% (236/307)	307	95% CI 71.8% to 81.2%	2026-08-24
Say-nothing rate (declines to bluff when the answer is not in the record)	100.0%	600	95% CI 99.4% to 100.0%	2026-08-25
Abstained on statute pinpoint	7.2% (22/307)	307	count, no interval reported	2026-08-24
Coaching quality (GW-14x, 0 to 8)	6.64 / 8	-	rubric mean, no interval reported	2026-08-24

Source files: hard-llm-level-llama4-scout-statute.json, hard-llm-level-llama4-scout-statute_rag.json, gw14x-ortier-llama4scout.json, hard-llm-level-llama4-scout-saynothing.json. Raw model name in source: meta-llama/llama-4-scout.