docs/how_to/installation/

[giscus[bot]]

docs/how_to/installation/

The LangChain ecosystem is split into different packages, which allow you to choose exactly which pieces of

https://python.langchain.com/docs/how_to/installation/

[eric-burel]

Hi, it would be nice to mention the "extras" syntax that is found in some tutorials for installing all dependencies at once with eg "pip install langchain[openai]". This was the first time I saw this syntax and it's not documented on this page.

[sjanaX01]

Why are there so many LangChain packages like langchain, langchain-core, and langchain-openai? It's very confusing for beginners. It's not clear which one is the main package and whether just one of them is enough to follow the documentation. This makes it harder to understand for someone new.

In tutorials, sometimes they use langchain-openai, and other times they use langchain. The way you select models is also different depending on which one you use. When you use langchain, the model selection process is different again.

So my main question is: If I only install langchain, is it enough to use all its features like a global package, or do I still need to install extra packages like langchain-openai? Just like the image I saw above.

If I just install the langchain and I want to use the GEMINI_API_KEY then how can use this just using the langchain package

[Sasha-BabyBird]

It would be nice if you ensured all the additional packages can be installed via conda. At the time, I can't find "langchain-ollama" in conda-forge.

[shashanksrajak]

As a beginner it is very difficult to navigate the installation and all the packages available. Modular design is one thing, but your docs are really confusing for new users on how to use these different packages. Someone should really work on this.

[Shingii]

Would like to ask if there could be more information for beginners on installing and usage. Documentation is rather confusing