A Rich-powered Python CLI to list, benchmark, and compare Ollama models across local and cloud endpoints. Measures time-to-first-token (TTFT) and tokens-per-second (TPS) with multi-prompt averaging, persists runs to a local SQLite database, and shows inline trend arrows when historical data exists. Results are exportable to JSON/CSV and auto-published to a live GitHub Pages dashboard via scheduled GitHub Actions, with filtering, sorting, and time-series charts. Clean six-module architecture; installable as a uv tool or via pip.

17 stars on GitHub; MIT licensed; live auto-updating dashboard at EndoTheDev.github.io/OMeter

Active — v0.4.0

A curated, opinionated "awesome list" of tools, web/desktop UIs, frameworks, libraries, and resources for Ollama — spanning coding agents, assistants, IDE integrations, chat/RAG apps, and model evaluation tools. Community-maintained with outside contributors submitting projects, and recognized on the awesome.re index.

474 stars, 59 forks on GitHub; community-contributed entries since 2024

Active

A modern web interface for Ollama, built with Nuxt 4, Nuxt UI v4, Pinia, and Tailwind. Features a streaming chat interface, model management (copy/rename/delete, search/sort), dark/light themes with a custom color picker, assistant thinking-block parsing, and a responsive layout. Ships with Docker and Docker Compose deployment and a Vitest test suite.

19 stars on GitHub; Vue/TypeScript, Dockerized, test-covered

Active — work in progress

A lightweight AI agent in ~300 lines of Python that chats with Ollama and uses tools — bash commands plus file read/write — with real-time streaming responses and a persistent interactive mode. Built on the native Ollama API with only httpx and python-dotenv, no extra SDKs. Supports both interactive and one-shot modes against local or cloud Ollama endpoints.

~300 lines of Python, no heavy dependencies

Active