Phi-3 Medium
MicrosoftLLMs
78.2
Performance
★ 4.2
Rating
312
Reviews
MultimodalMediumText GenerationOpen SourceSmall
About
Microsoft's highly capable 14B parameter model designed for on-device and edge deployment with surprisingly strong benchmarks.
Strengths
Remarkably capable for 14B parameters, often compared to Mixtral 8x7B and GPT-3.5. Excellent for local/edge deployment on consumer hardware. Strong on knowledge (MMLU) relative to size. MIT license.
Specifications
- Context window
- 128,000
- Parameters
- 14B
Pricing
- Input cost
- Free
- Output cost
- Free
Open-weight (MIT license). Free to download. Runs on consumer GPUs.
Speed & Latency
- 120
- tokens/sec
- 150ms
- time to first token
Available On
HuggingFaceAzure AIOllamaONNX Runtime
Features
streamingsystem messages
Performance Trend
Benchmark score trends over time for the top 5 benchmarks.
Loading history...
Benchmarks
Scores from various benchmark tests; higher is better.
| Test | Score | Percentile | Source |
|---|---|---|---|
| BigBench Hard | 68.0 | p78 | seed |
| DROP | 70.0 | p79 | seed |
| GSM8K | 78.0 | p84 | seed |
| HumanEval | 70.0 | p78 |