txtai
txtai is a Python framework that helps developers search documents by meaning and build question-answering apps, workflows and agents on top of language models.
What it is good for
- Searching a pile of documents by meaning instead of keywords
- Building a chat app that answers questions from your own data
- Chaining several AI models together into one automated workflow
- Building an agent that solves multi-step problems on its own
At a glance
- Who it is for
- Setup difficulty
- Developer ยท Requires programming
Level 5 of 5. You need to write code to use it.
- Cost
- Free and open source
- Type
- Framework or SDK
- Tasks
- Works with
- Hugging Face, llama.cpp, OpenAI, Claude, AWS Bedrock, Docker
How to get started with txtai
- Install txtai with pip or run it with Docker.
- Write a few lines of Python that import txtai and create an embeddings index.
- Or write an
app.ymlconfig file and start the built-in API with uvicorn. - Open the example notebooks to learn each feature step by step.
From the project's own documentation. If a step does not match what you see, the official pages have the latest instructions.
Before you try it
- Requires Python 3.11 or newer.
- You need to write code or edit config files to use it.
- Agents are built on smolagents and work with several model providers, which may need their own accounts.
Momentum
- Rank this week
- #37 in builder tools
- Best rank
- #37 week 46, 2026
- Stars gained last week
- +215
- GitHub stars
- 13,408
GitHub numbers as of Nov 16, 2026. Updated weekly.
Ranking history
Official links
Questions
- Is txtai free?
- Yes. It is free and open source (Apache-2.0 license).
- Do I need to know how to code to use txtai?
- Yes. You need to write code to use it.
- What does txtai work with?
- According to its own documentation, it works with Hugging Face, llama.cpp, OpenAI, Claude, AWS Bedrock, and Docker.
- How do I get started with txtai?
- 1. Install txtai with pip or run it with Docker. 2. Write a few lines of Python that import txtai and create an embeddings index. 3. Or write an app.yml config file and start the built-in API with uvicorn. 4. Open the example notebooks to learn each feature step by step.
- What should I know before trying txtai?
- Requires Python 3.11 or newer. You need to write code or edit config files to use it. Agents are built on smolagents and work with several model providers, which may need their own accounts.
- What is txtai good for?
- Searching a pile of documents by meaning instead of keywords, building a chat app that answers questions from your own data, chaining several AI models together into one automated workflow, and building an agent that solves multi-step problems on its own.
Similar agents to txtai
- LlamaIndex
Focuses on searching and answering over your own documents, with less emphasis on built-in vector database and graph storage.
- Haystack
Aimed at production LLM apps and agent workflows, without txtai's own embeddings database at the center.
- PageIndex
Answers questions over long documents using a tree index instead of a vector database.
- CocoIndex
Concentrates on keeping data fresh for agents by reprocessing only what changed, not on search itself.
- Yuxi
A self-hosted platform with shared permissions for teams, rather than a code library.
- Onyx
Onyx is an open source AI chat platform that searches your team's apps and documents to answer questions for teams.