Model components / Fine-tuned / chat models

Phi-4-mini-instruct

Compact instruction-tuned small language model (3.8B) that punches well above its weight on coding benchmarks. Part of Microsoft's Phi-4 family, which expanded across 2025–early 2026 with Phi-4-reasoning, Phi-4-multimodal, and Phi-4-Reasoning-Vision-15B (March 2026); no Phi-5 has shipped as of May 2026. 1.6M monthly downloads make Phi-4-mini one of the most popular small models for local coding.

Phi-4-mini-instruct, 3.8B dense, 128k context, 200k vocab, GQA; released Feb 2025 under MIT. SFT/DPO-aligned small instruct model. Verified live on HF June 2026.

Openness

3 high confidence

3.0

weights: open(MIT, on HF)
data: closed(synthetic-heavy training data undisclosed)
code: closed(no training pipeline)
license: MIT(OSI permissive)

MIT-licensed open weights; training/post-training data and code not released -> open_weights.

https://huggingface.co/microsoft/Phi-4-mini-instruct recorded 2026-06-04
live card; license=MIT; 3.8B instruct; 1,537,561 downloads last month

Adoption

4 high confidence

4.0

~1.54M downloads last month on HF; widely distributed via Azure AI Foundry / Ollama for on-device + local inference. Solidly in the 1-10M band.

https://huggingface.co/microsoft/Phi-4-mini-instruct recorded 2026-06-04
1,537,561 downloads last month

Capability

2 high confidence

2.0

Strong-for-size 3.8B with good math (GSM8K 88.6) but low GPQA (25.2) and overall well below 2026 frontier chat models; small-model tier on a frontier-anchored scale.

https://huggingface.co/microsoft/Phi-4-mini-instruct recorded 2026-06-04
benchmark table: MMLU-Pro 52.8, GPQA 25.2, GSM8K 88.6, overall 63.5

Unchanged since 2026-07-29 (last edited, not re-checked)