Microsoft
Microsoft's AI research teams release a wide variety of open source models, including popular text embedding base models.
Microsoft models
About Microsoft
Microsoft’s AI research divisions create a variety of models from LLMs to speech synthesis to text embedding — over 200 published models. One of their areas of focus is maintaining performance for smaller, more efficient LLMs such as the Phi family of models.
Microsoft is closely partnered with OpenAI and backs open source research teams like WizardLM.
What license does Microsoft release models under?
Microsoft uses a variety of licenses on a model-by-model basis. Always check the license of the model you intend to use before deployment.
Microsoft’s large language model families like Orca 2 and Phi 2 are released under a special Microsoft Research License that does not allow for commercial use.
However, many of their utility models, such as MiniLM text embedding models, are released under the MIT license, which allows for commercial use.
