Compare

Llama 4 Behemoth vs Qwen 3 Max

Meta vs Qwen. Specs, benchmarks, and real per-task cost — all in one page.

Verdict

Llama 4 Behemoth leads on LMSYS ELO (1342 vs 1295). Qwen 3 Max is ~1.4× cheaper on a 3:1 input:output blend.

Meta

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

Best open-weights modelRuns anywhereTransparent training

Qwen

Alibaba’s flagship. Strongest non-English coverage of any open model; dominant in Asian markets.

MultilingualOpen weightsCompetitive coding

Pricing

Input / 1M$2.50$1.60

Output / 1M$8.00$6.40

Context256K128K

Max output16K16K

LicenseOpen weightsOpen weights

Released2025-12-102025-11-20

Benchmarks

LMSYS ELO1342.01295.0

MMLU Pro87.184.0

HumanEval89.387.2

SWE-Bench52.4—

MATH81.585.1

GPQA56.2—

MMMU—71.8

IFEval88.085.5

Per-task cost

Summarize a 1-hour meeting transcript$0.041$0.027

Review a 500-line pull request$0.036$0.026

Answer a customer support ticket$0.014$0.0099

Extract structured data from a resume$0.021$0.014

Debug a stack trace with context$0.031$0.022

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

Add up to four models, tweak tasks live.