Ollama
3 Tools · 4 Intel
Ollama is an open-source company founded in 2023 by Jeffrey Morgan and Michael Chiang, University of Waterloo alumni and Y Combinator W21 graduates. Morgan brings a developer-tooling background from Docker, which shaped the company's philosophy of packaging large language models into something any developer can run on their own machine without cloud dependencies.
Tools
Official Python library for running, managing, and chatting with locally served models through Ollama.
JavaScript library for running, managing, and chatting with locally served Ollama models from Node or the browser.
Intel
Ollama v0.40.7 adds the Perplexity pplx-decider 27B model, fixes OpenAI-compatible Responses conversations across thinking and non-thinking models, orders streamed tool-call events correctly, and exposes renderer, parser and template source metadata in the API.
Ollama 0.40.9 adds Phonon speech generation on Apple Silicon with an OpenAI-compatible audio endpoint, d3 classification and scoring models in four sizes, and Flint on Linux and Windows. Streaming requests now report generation failures, and launch passes cloud model context lengths to coding tools.
Ollama's latest release adds Flint, Recursive's open-weight 40B MoE coding model, runnable locally with `ollama run flint`. It reports 54.9% on Terminal-Bench 2.1 and 74.2% on SWE-bench Verified with a 262K context window, currently macOS-only.
Ollama v0.40.10 adds NVIDIA's Nemotron 3.5 Bolt E4B, a Gemma 4-based vision-language model with 128K context, tool use and a DFlash speculative-decoding draft, runnable locally via a single ollama run command.