← LLM Models#25Open SourceView on HF →
Qwen 2.5 72B
Alibaba/Qwen
Alibaba's open powerhouse. Tops coding and math benchmarks at 72B.
Official page →📖Context128K tokens
🚀Speed80 tok/s
💵Input price-
💸Output price-
🧠Parameters72B
Benchmarks
Gray bar = dataset averageMMLU57 academic subjects
86.1%avg 81.8
GPQA DiamondPhD-level science questions
49.0%avg 50.4
HumanEvalPython code generation
86.5%avg 79.9
SWE-Bench VerifiedReal GitHub engineering tasks
23.7%avg 32.4
Composite ScoreScore = MMLU×20% + GPQA×30% + HumanEval×25% + SWE-Bench×25%59.5
HF Open LLM Leaderboard
IFEval · BBH · MATH · GPQA · MuSR · MMLU-Pro
Average
48.0
IFEvalInstruction following
86.4%avg 79.9
BBHBig Bench Hard
61.9%avg 48.7
MATH Lvl 5Competition mathematics
59.8%avg 37.8
GPQAGraduate-level science Q&A
16.7%avg 12.9
MuSRMultistep soft reasoning
11.7%avg 10.7
MMLU-ProProfessional knowledge
51.4%avg 40.0
More from Alibaba/Qwen
Details
- Provider
- Alibaba/Qwen
- Parameters
- 72B
- Context
- 128K tokens
- Speed
- 80 tok/s
- Open Source
- Yes
- License
- Open (varies)
- Released
- -