Compare

Llama 4 Scout vs DeepSeek R2

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

Verdict

Llama 4 Scout is ~2.1× cheaper on a 3:1 input:output blend.

Meta

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

Single-GPU deployableDirt cheap on aggregatorsStrong fine-tuning base

DeepSeek

DeepSeek’s reasoning variant. Competes with o4 on math at a fraction of the cost.

Frontier math at open-weights pricingLong CoT traces

Pricing

Input / 1M$0.35$0.55

Output / 1M$0.80$2.19

Context128K128K

Max output8K64K

LicenseOpen weightsOpen weights

Released2025-12-102026-01-15

Benchmarks

LMSYS ELO1265.0—

MMLU Pro77.287.8

HumanEval82.490.1

MATH72.594.1

GPQA—71.2

IFEval84.1—

Per-task cost

Summarize a 1-hour meeting transcript$0.0057$0.0093

Review a 500-line pull request$0.0044$0.0088

Answer a customer support ticket$0.0017$0.0034

Extract structured data from a resume$0.0026$0.0049

Debug a stack trace with context$0.0037$0.0077

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

Add up to four models, tweak tasks live.