Microsoft

Phi-3-medium

Compact yet capable agent for on-device and edge applications.

Strong performance for its sizeOptimized for smaller, faster inferenceGood reasoning and instruction followingSuitable for on-device deployment
Today's score
87.0
Try Phi-3-medium

Where it ranks today

Best for / Not great for

Best for
  • Mobile applications
  • Edge computing agents
  • Offline AI assistants
  • Lightweight development tasks
Not great for
  • Extremely complex, multi-step reasoning
  • Processing very large documents
  • Cutting-edge multimodal capabilities

Why it ranks here

Microsoft's Phi-3-medium is gaining significant traction for its impressive performance in a compact size, making it ideal for agentic applications on edge devices and mobile platforms. Its efficiency and capable reasoning open new possibilities for on-device AI.

30-day trend

Score breakdown

Search trends85
Benchmarks88
Developer buzz90
News mentions87

Pricing

API: $0.00 in · $0.00 out per 1M tokens · Consumer: $0.00/mo

Pricing plans

Popular
Azure AI (Phi-3)
Access Phi-3 models on Azure.
$0 /usage
  • Phi-3-medium access
  • Optimized inference
  • Pay-as-you-go pricing
  • Integration with Azure services
Get started on Azure
Open Source Model
Download and run Phi-3 locally.
Free
  • Phi-3-medium weights
  • Permissive license
  • Requires self-hosting
  • Community support
Download model
Compare with another modelHow is this score calculated? →Snapshot 2026-06-25