Compare

Llama 4 Behemoth vs Claude Haiku 4.5

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

Verdict

Llama 4 Behemoth leads on LMSYS ELO (1342 vs 1305). Claude Haiku 4.5 is ~2.4× cheaper on a 3:1 input:output blend. Llama 4 Behemoth is open-weights, the other is proprietary.

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

Anthropic

Claude Haiku 4.5

Fast, cheap, and smart enough for most routing and extraction tasks. Great base model for subagents.

Sub-second latency10x cheaper than OpusStrong structured output

Pricing

Input / 1M$2.50$0.80

Output / 1M$8.00$4.00

Context256K200K

Max output16K16K

LicenseOpen weightsProprietary

Released2025-12-102025-10-01

Benchmarks

LMSYS ELO1342.01305.0

MMLU Pro87.181.4

HumanEval89.387.0

SWE-Bench52.449.2

MATH81.576.8

GPQA56.2—

IFEval88.087.3

Per-task cost

Summarize a 1-hour meeting transcript$0.041$0.014

Review a 500-line pull request$0.036$0.014

Answer a customer support ticket$0.014$0.0056

Extract structured data from a resume$0.021$0.0080

Debug a stack trace with context$0.031$0.013

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

Tool

Interactive compare →

Add up to four models, tweak tasks live.