Compare
Llama 4 Behemoth vs Llama 4 Scout
Meta vs Meta. Specs, benchmarks, and real per-task cost — all in one page.
Verdict
Llama 4 Behemoth leads on LMSYS ELO (1342 vs 1265). Llama 4 Scout is ~8.4× cheaper on a 3:1 input:output blend.
Meta
Llama 4 BehemothMeta’s open-weights flagship. 405B params, fully open license, runs on every major inference provider.
Best open-weights modelRuns anywhereTransparent training
Meta
Llama 4 ScoutThe 70B workhorse open-weight. Fits on a single H100, beloved by self-hosters.
Single-GPU deployableDirt cheap on aggregatorsStrong fine-tuning base
Pricing
Input / 1M$2.50$0.35
Output / 1M$8.00$0.80
Context256K128K
Max output16K8K
LicenseOpen weightsOpen weights
Released2025-12-102025-12-10
Benchmarks
LMSYS ELO1342.01265.0
MMLU Pro87.177.2
HumanEval89.382.4
SWE-Bench52.4—
MATH81.572.5
GPQA56.2—
IFEval88.084.1
Per-task cost
Summarize a 1-hour meeting transcript$0.041$0.0057
Review a 500-line pull request$0.036$0.0044
Answer a customer support ticket$0.014$0.0017
Extract structured data from a resume$0.021$0.0026
Debug a stack trace with context$0.031$0.0037
Per-call cost using published token counts for each task. Real-world prompts vary.