Inference Index
Compare

Llama 4 Scout vs Llama 4 Behemoth

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 Scout

The 70B workhorse open-weight. Fits on a single H100, beloved by self-hosters.

Single-GPU deployableDirt cheap on aggregatorsStrong fine-tuning base
Meta
Llama 4 Behemoth

Meta’s open-weights flagship. 405B params, fully open license, runs on every major inference provider.

Best open-weights modelRuns anywhereTransparent training
Pricing
Input / 1M$0.35$2.50
Output / 1M$0.80$8.00
Context128K256K
Max output8K16K
LicenseOpen weightsOpen weights
Released2025-12-102025-12-10
Benchmarks
LMSYS ELO1265.01342.0
MMLU Pro77.287.1
HumanEval82.489.3
SWE-Bench52.4
MATH72.581.5
GPQA56.2
IFEval84.188.0
Per-task cost
Summarize a 1-hour meeting transcript$0.0057$0.041
Review a 500-line pull request$0.0044$0.036
Answer a customer support ticket$0.0017$0.014
Extract structured data from a resume$0.0026$0.021
Debug a stack trace with context$0.0037$0.031

Per-call cost using published token counts for each task. Real-world prompts vary.