Benchmark

Meta Llama — reported benchmarks

Meta · source page ↗ · last checked Oct 5, 2026, 12:03 PM

Reported benchmarks · Llama 4 Maverick

captured Oct 5, 2026, 12:03 PM
BenchmarkScore
MMLU Pro80.5% · instruction-tuned, 0-shot
GPQA Diamond69.8% · instruction-tuned, 0-shot
MMMU73.4% · instruction-tuned, image reasoning, 0-shot
MathVista73.7% · instruction-tuned, 0-shot
ChartQA90.0% · instruction-tuned, 0-shot relaxed accuracy
DocVQA94.4% · instruction-tuned, 0-shot anls metric
LiveCodeBench43.4% · instruction-tuned, 0-shot pass@1, 10.01.2024-02.01.2025
MATH61.2% · pre-trained, 4-shot em_maj1@1
MBPP77.6% · pre-trained, 3-shot pass@1
Multilingual MMLU84.6% · instruction-tuned
MTOB (half-book)54.0/46.4 chrF · instruction-tuned, eng->kgv/kgv->eng long context

Vendor-reported via automated web search — not independently verified. See the cited matrix on /models.

Change history