oterm is the terminal client for LLMs: a Python TUI that connects one terminal interface to Ollama, OpenAI, Anthropic, and any pydantic-ai-supported provider, aimed at developers and terminal-centric teams who would rather chat with models from the command line than from a browser.

What it is

oterm is a Python application, released under the MIT licence, that provides a terminal user interface for chatting with large language models. Its topics describe it as a chatbot and a TUI built with Textual, filed under AI & Machine Learning / AI Interaction & Interfaces, and it targets the provider ecosystems listed alongside it: anthropic, ollama, openai, and pydantic-ai. The project lives on GitHub, where it has 2,442 stars and 138 forks against 4 open issues, and it keeps a documentation site at https://ggozad.github.io/oterm/ alongside a CHANGELOG.md.

The problem it solves is fragmentation. Ollama, OpenAI, and Anthropic each offer their own chat interfaces, and pydantic-ai exposes a further range of providers, so a developer working in a shell has to leave it for a browser or juggle several clients depending on which model is wanted. oterm replaces that per-provider, browser-bound arrangement with a single terminal client that reaches all of them, keeping the conversation inside the environment where the work is already happening. Its topic list also includes mcp, the Model Context Protocol.

Key capabilities

• One terminal client covering Ollama, OpenAI, and Anthropic, so several providers are reachable without switching tools.
• Support for any provider that pydantic-ai supports, extending coverage beyond the three named providers.
• A terminal user interface built with Textual, reflected in the textual and tui topics, rather than a web front end.
• Installable and runnable straight from the published package with the documented uvx oterm command.
• MCP listed among the topics, indicating integration with the Model Context Protocol.
• MIT-licensed Python source recorded in the LICENSE file, with documentation at the project homepage and changes tracked in CHANGELOG.md.

Who uses it and how

• Individual developers running Ollama locally who want to talk to local models from the same terminal they code in.
• Teams mixing local and hosted models, using one client for an Ollama instance and for OpenAI or Anthropic API access.
• Users of pydantic-ai providers beyond the three named, who inherit that coverage through oterm.
• Terminal-centric and remote workflows, where a TUI works in a shell that has no graphical chat application.
• Evaluators and contributors, drawn by a project with 2,442 stars, 138 forks, and 4 open issues.

Getting started

The README's installation line is uvx oterm, and it points to https://ggozad.github.io/oterm/ for full install methods, configuration, and usage.

How it compares

No paid products that oterm replaces appear in the facts provided, and no comparable terminal client is named either, so it stands alone in this registry; the other projects named alongside it are providers it connects to rather than alternatives it competes with. A fair comparison is therefore against each provider's own first-party interface, which is the arrangement oterm is designed to replace.

When to use it — and when not to

Adopting oterm means supplying your own model access: a local Ollama instance or API credentials for OpenAI, Anthropic, or another pydantic-ai provider, since no hosted service is described. It fits people who are at home in a terminal, and it is a poor choice for anyone who wants a graphical or browser-based chat interface or a fully managed service with no local setup. The README itself is sparse, offering the uvx oterm command and a link to external documentation, so configuration and daily usage have to be learned from the documentation site, and no released version or date is given beyond the CHANGELOG reference.

uvx oterm