Phi-3 Medium
Microsoft
Microsoft's efficient Small Language Model (SLM) with 14B parameters. Released May 2024 as part of Phi-3 family alongside Mini (3.8B) and Small (7B). Available in two context lengths: 128K and 4K. Outperforms Gemini 1.0 Pro and GPT-3.5 Turbo despite significantly smaller size. Demonstrates strong reasoning and logic capabilities making it suitable for analytical tasks beyond traditional language tasks. Optimized variants available with ONNX Runtime and DirectML supporting wide range of devices including mobile and web deployments. Available as NVIDIA NIM inference microservices with standard API. Represents Microsoft's 'small but mighty' SLM approach.
Strengths
- Outperforms much larger models - beats Gemini 1.0 Pro and GPT-3.5 Turbo
- 128K context window impressive for 14B parameter model
- Strong reasoning and logic - suitable for analytical tasks
- Optimized for edge deployment - mobile, web, IoT devices
- ONNX Runtime and DirectML variants for wide platform support
- MIT license enables unrestricted commercial use
- NVIDIA NIM microservices for standardized deployment
Caveats
- 14B size still limits capabilities vs 70B+ flagship models
- 4K max output tokens standard for size class
- Text-only (no vision or multimodal)
- Smaller model may struggle with highly specialized domains
- Less ecosystem maturity than GPT or Claude families
- Vendor
- Microsoft
- Context Window
- 128,000 tokens
- Max Output
- 4,096 tokens
- License
- MIT
- Release Date
- 2024-05-01
- Modalities
- text
Capabilities
Resources
Pricing
Open source (MIT) - free. Optimized for edge deployment. API providers typically $0.20-$0.50 per 1M input tokens