txtai
neuml.github.io
- Free plan
- Windows
- macOS
- Linux
- API
- Self-hosted
What txtai is
txtai is a free, open-source framework for semantic search, language-model workflows, and LLM orchestration. Its embeddings database combines vector indexes with graph and relational data, supporting vector search, SQL, object storage, topic modeling, graph analysis, and multimodal indexing. It can create embeddings from text, documents, audio, images, and video. Language-model pipelines cover tasks such as question answering, transcription, translation, and summarization; workflows join pipelines with business logic, while agents can coordinate those components autonomously. txtai offers Web and Model Context Protocol APIs and bindings for JavaScript, Java, Rust, and Go. It runs locally, on servers or clusters, and in serverless environments. Installation is via pip or Docker, and Python 3.11 or later is supported. The project is available under the Apache 2.0 license at no cost. Its API is open over HTTP by default; documentation describes HTTPS and token authorization options. Safeopen is enabled by default for API and application runs.
Who it suits
txtai suits developers and teams building semantic search, retrieval, or language-model workflows who want to run locally or self-host. Its documented language and model integrations may also suit projects using Hugging Face, llama.cpp, OpenAI, Claude, or AWS Bedrock through LiteLLM.
Good to know, and what to watch
- Free under the Apache 2.0 license.
- Creates embeddings from text, documents, audio, images, and video.
- Includes Web and Model Context Protocol APIs.
- Runs locally, on servers, clusters, or serverless platforms.
- Requires Python 3.11 or later for the documented install.
- API is open over HTTP by default.
- Hosted txtai.cloud is described as planned, with no plan pricing listed.
Plans
Every plan and pricetxtai
Open source
Getting started
- Open the txtai documentation website
- Install txtai with pip from PyPI
- Use Docker for a CPU or GPU installation if preferred
- Run it locally, on a server, cluster, or supported serverless environment
- Configure HTTPS, API-token authorization, and safeopen settings for API deployments
The verdict
txtai brings search, data indexing, and language-model workflows into one framework, with local and varied deployment options. Teams should account for its Python requirement and configure API security beyond the default HTTP setup.
Questions people ask
Is txtai free?
Yes. txtai is open source under the Apache 2.0 license, and the listed txtai plan costs nothing.
How do I install txtai?
The documented routes are pip installation from PyPI or Docker images for CPU and GPU environments. Python 3.11 or later is supported.
Can txtai run locally?
Yes. txtai can run locally, on a server, on a cluster, or in documented serverless environments.
Which APIs and language bindings are available?
txtai provides Web and Model Context Protocol APIs, with bindings for JavaScript, Java, Rust, and Go.
Which models can txtai connect to?
The documentation describes agent support for Hugging Face and llama.cpp models, plus OpenAI, Claude, and AWS Bedrock through LiteLLM.
Is txtai open source?
Yes. It is released under the Apache 2.0 license.
What the maker publishes
- APIs and bindings
- txtai provides Web and Model Context Protocol APIs, with bindings for JavaScript, Java, Rust, and Go. source
- API security
- The API is open over HTTP by default; the security guide recommends enabling HTTPS and documents API-token authorization. source
- Deployment
- txtai API instances and workflows can run locally, on a server, on a cluster, or serverless; documented serverless options include AWS Lambda, Google Cloud Functions, Azure Cloud Functions, and Kubernetes with Knative. source
- Embeddings database
- Its embeddings database combines sparse and dense vector indexes, graph networks, and relational databases. source
- Features
- It supports vector search, multimodal embeddings, language-model pipelines, workflows, and autonomous agents. source
- Formats
- It can create embeddings for text, documents, audio, images, and video. source
- Hosted offering
- The official pages describe txtai.cloud as a planned effort for hosted txtai applications, without listing plan pricing. source
- Install and runtime
- The documented install is via pip and PyPI, Python 3.11+ is supported, and Docker images are available for CPU and GPU installs. source
- Integrations
- The documentation describes agent support for Hugging Face and llama.cpp models, and OpenAI, Claude, and AWS Bedrock through LiteLLM. source
- License
- txtai is open-source under the Apache 2.0 license. source
- License and price
- txtai is open-source under the Apache 2.0 license; the official pages reviewed describe pip installation and self-hosting and do not list a software price. source
- Local operation
- txtai can run locally, and its documentation says this avoids shipping data to disparate remote services. source
- Local use
- The maker says txtai can run locally, avoiding the need to send data to separate remote services. source
- Media support
- txtai can create embeddings for text, documents, audio, images, and video. source
- Operating systems
- The installation guide provides environment-specific prerequisites for Linux, macOS, and Windows. source
- Pipelines
- Language model pipelines can handle prompts, question answering, labeling, transcription, translation, and summarization. source
- Product
- txtai is an all-in-one AI framework for semantic search, LLM orchestration, and language model workflows. source
- Purpose
- txtai is an all-in-one AI framework for semantic search, LLM orchestration, and language model workflows. source
- Requirements
- The installation guide says Python 3.11 or later is supported. source
- Safe file and URL access
- The safeopen option, enabled by default for the API and applications, restricts local files to a configured directory and URLs to public addresses. source
- Search and data
- Its embeddings database combines vector indexes, graph networks, and relational databases, and supports vector search with SQL, object storage, topic modeling, graph analysis, and multimodal indexing. source
- Security
- The API is open over HTTP by default; the documentation describes HTTPS options, API-token authorization, and a safeopen option enabled by default for API and application runs. source
- Support
- NeuML offers implementation guidance, development support, and AI consulting, with meetings or email as contact options. source
- Workflows and agents
- Workflows join pipelines and business logic, while agents connect embeddings, pipelines, workflows, and other agents to solve problems autonomously. source
Checked against the maker's pages on 5 Oct 2026.