Model Comparison

Gemma 2 9B

Google

Gemma 2 9B is a language model developed by Google. It achieves strong performance with an average score of 64.6% across 16 benchmarks. It excels particularly in ARC-E (88.0%), BoolQ (84.2%), HellaSwag (81.9%). The model shows particular specialization in reasoning tasks with an average performance of 79.7%. It's licensed for commercial use, making it suitable for enterprise applications. Released in 2024, it represents Google's latest advancement in AI technology.

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.