Qwen2 Instruct 72B

Alibaba

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$0 / $0

in / out

Context

—

Avg score

35.3

6 benchmarks

AA Intelligence

6.0

index

Weights

Open

Released

Jun 7, 2024

About

Qwen2 Instruct 72B is a open-weights model in the Qwen2 family from Alibaba. Benchmarked on 6 evals, averaging 35.3.

Benchmark scores · 6

avg 35.3 — higher is better

MATHartificial-analysis · 2026-07

70.1

MMLU-Proartificial-analysis · 2026-07

62.2

GPQA Diamondartificial-analysis · 2026-07

37.1

SciCodeartificial-analysis · 2026-07

22.9

LiveCodeBenchartificial-analysis · 2026-07

15.9

Humanity's Last Examartificial-analysis · 2026-07

3.7

Details

PublisherAlibaba

FamilyQwen2

Parameters—

License—

WeightsOpen

ReleasedJun 7, 2024

History

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0/M