Microsoft
Microsoft's 14B Phi-4 model — exceptional reasoning and STEM performance that outperforms many larger models.
Phi-4 Prompt — rank #70
Microsoft's compact 3.8B Phi-4 Mini — strong reasoning and STEM performance in a small, edge-friendly package with 128K context.
Phi-4 Mini proxy (GAIA era)
Phi-4 Mini — Phi-4 Prompt family
