Phi 4Update Logs
View product update logs, including latest features and improvements.
Chat with Phi 4 on phi4.lol
phi4.lol now serves the phi-4 open model behind a full chat experience, so you do not need to install anything to try it.
What is new
- Chat with Phi 4 at phi4.lol/chat
- Sign in with Google and claim starter credits
- Every generation draws from your credits, top up anytime on the pricing page
- phi-4 models are available in GGUF formats, from 2 GB up to 8 GB of RAM locally
About the model
Phi-4 is designed to accelerate research on language models, for use as a building block for generative AI powered features. It provides uses for general purpose AI systems and applications (primarily in English) which require memory/compute constrained environments, latency bound scenarios, and reasoning and logic.
Phi-4 is provided under the MIT license.
Phi 4 (phi-4) Released
phi-4 is a state-of-the-art open model built upon a blend of synthetic datasets, data from filtered public domain websites, and acquired academic books and Q&A datasets. The goal of this approach was to ensure that small capable models were trained with data focused on high quality and advanced reasoning.
Highlights
- 14B parameter open model from Microsoft, released under the MIT license
- Trained on synthetic datasets, filtered public domain websites, and academic books and Q&A datasets
- Supervised fine-tuning and direct preference optimization for precise instruction adherence and robust safety
- 84.8 on MMLU, 56.1 on GPQA, 80.6 on MGSM, and 82.6 on HumanEval
Run it locally
phi-4 models are available in GGUF formats. To run the smallest phi-4 you need at least 2 GB of RAM; the largest one may require up to 8 GB.
# Browse phi-4 GGUF builds
open https://lmstudio.ai/models/phi-4Technical report: https://arxiv.org/pdf/2412.08905