Find your data from at (alpha version). Join the discord for live updates:

Updating Chainlit

Begin the migration by updating Chainlit to the latest version:

pip install --upgrade chainlit

The documentation has been entirely updated, click on the links below to see the new APIs!


Previously, messages were both used for chat interactions and display intermediate steps.

Intermediary steps are more than just messages, they have an input and an output, a start and an end.

For this reason, we decided to remove the parent_id, prompt and indent fields from the cl.Message class and introduce a new concept: cl.Step.

Finally, the disable_human_feedback field has been renamed to disable_feedback.


The Chainlit Prompt class was mixing fields for both chat and completion LLMs.

To make it clearer, it has been split into two classes: ChatGeneration and CompletionGeneration.

Generations are intended to be used with Step.


The AppUser class has been renamed to User. The username field has been renamed to identifier to emphasize that it has to be a unique identifier.

The other fields such as provider have been removed. A new metadata field has been added instead. It is a dictionary that can be used to store any information about the user.


The response of AskUserMessage has been slightly modified. The content field has been renamed to output.

File Upload

File upload has been reworked to remove the size limit and vastly improve performance. The APIs stay the same with one exception.

The file objects returned by AskFileMessage and multi modal messages no longer have the content of the file. Instead they have a path field that can be used to manipulate the file.

Chainlit Cloud

Chainlit Cloud has been replaced by our entirely new platform called Literal AI.

Literal AI not only works out of the box with Chainlit apps but also with any other LLM app. Your historical data has been migrated automatically.

In your Chainlit app, replace your API key definition with:



Literal AI brings a lot of new features such as:

  • User analytics
  • Chat analytics
  • Detailed traces
  • Python & TypeScript SDKs to interact with your data

and more!