Phi 4 Mini Instruct from Microsoft is a lightweight open model with 128K context, synthetic data training, and strong instruction-following focus.
Capabilities, design details, and architectural traits
Phi 4 Mini Instruct is Microsoft’s lightweight open model in the Phi-4 family. Official material describes it as built from synthetic data and filtered public websites, with a focus on high-quality reasoning and instruction following.
| Aspect | Official Characteristic |
|---|---|
| Model family | Part of Microsoft’s Phi-4 line. |
| Training data | Built on synthetic data and filtered public web data. |
| Context strategy | Supports a 128K-token context window. |
| Optimization focus | Emphasizes reasoning and precise instruction adherence. |
| Safety design | Documentation highlights safety-focused fine-tuning. |
The model is documented as a compact open model rather than a large frontier system. Its positioning centers on high-quality data curation and instruction tuning, with a clear emphasis on reasoning-oriented behavior. Official sources also present it as a lightweight option for deployments that need long context and lower resource usage.
Independent evaluations · Artificial Analysis
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Microsoft | Anthropic | Anthropic |
| Release Date | February 26, 2024 | July 24, 2026 | June 9, 2026 |
| Knowledge Cutoff | - | May 2026 | - |
| Context & Limits | |||
| Context Window | 131K | 1M Best Context Window | 1M Best Context Window |
| Pricing (per 1M tokens) | |||
| Input Pricing | Free Best Input Pricing | $5 | $10 |
| Output Pricing | Free Best Output Pricing | $25 | $50 |
| Modalities | |||
| Inputs | text | textimage | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 5.7 | 63.1 Best Intelligence Index | 62.1 |
| Coding Index | 3.8 | 78.0 Best Coding Index | 76.5 |
| Agentic Index | 0.3 | 59.2 Best Agentic Index | 56.6 |
Graduate-level reasoning and expert Q&A evaluation.
Extremely difficult logical reasoning and knowledge.
Logical reasoning over long context windows.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
