Langflow

What it is

Langflow is an open-source, low-code builder for AI agents and retrieval-augmented generation (RAG) applications. You assemble a flow on a canvas from components (chat input, models, tools, vector stores, outputs), test it in a built-in Playground, and then serve it as an API or as tools for an MCP server. Langflow was created by Logspace, which DataStax acquired in 2024. IBM completed its acquisition of DataStax in May 2025, and IBM now sells enterprise support for the project. The Python package is published under the MIT licence.

Key features

• Visual canvas plus code: drag-and-drop components grouped as Input & Output, Data Sources, Models & Agents, LLM Operations, Processing and Flow Control. Every component is Python, so you can edit one or write a custom component.
• Agents: an Agent component takes a language model, instructions and a set of tools, and other components or whole flows can be attached to it as tools.
• MCP in both directions: each project can be exposed as an MCP server whose tools are its flows, and flows can call external MCP servers as a client.
• Deploy as an API: flows run over a REST API, with code snippets for cURL, Python and JavaScript.
• Providers as bundles: since version 1.12 (September 2026), the default install includes a curated set of providers and others are installed as extension packages. Version 1.12 also added OpenTelemetry traces, metrics and logs.

Platforms

Langflow Desktop runs on macOS 13 or later and Windows, and the download requires a short sign-up form. You can also run the langflowai/langflow Docker image or install the Python package (uv pip install langflow, Python 3.10 to 3.14). The docs list a dual-core CPU and 2 GB of RAM as the minimum for the package. The hosted DataStax Langflow service on Astra was a tech preview and was shut down on 9 April 2026, and DataStax points users to the open-source version instead.

Who it's for

Developers and technical teams who want to prototype agents and RAG pipelines visually without giving up access to the underlying Python, and who are comfortable hosting it themselves.

Pricing

Langflow itself is free and open source, with no paid tier of the software. IBM offers IBM Elite Support for Langflow, a paid support subscription for teams running their own open-source deployments. Its price is not published, and as of October 2026 IBM asks prospective customers to get in touch.