Model Comparison

Phi 4 Mini

Microsoft

Phi 4 Mini is a language model developed by Microsoft. It achieves strong performance with an average score of 65.4% across 17 benchmarks. It excels particularly in GSM8k (88.6%), ARC-C (83.7%), BoolQ (81.2%). It's licensed for commercial use, making it suitable for enterprise applications. Released in 2025, it represents Microsoft's latest advancement in AI technology.

QwQ-32B-Preview

Alibaba

QwQ-32B-Preview is a language model developed by Alibaba. It achieves strong performance with an average score of 64.0% across 4 benchmarks. It excels particularly in MATH-500 (90.6%), GPQA (65.2%), AIME 2024 (50.0%). The model is available through 4 API providers. It's licensed for commercial use, making it suitable for enterprise applications. Released in 2024, it represents Alibaba's latest advancement in AI technology.