Does this require a real OpenAI API key?

No. While the OpenAI Python library requires an API key string to initialize, Ollama ignores the value. You can use a placeholder like 'ollama' for all local requests.

Which OpenAI endpoints are supported through Ollama?

The compatibility layer supports /v1/models for listing, /v1/completions for text generation, /v1/chat/completions for chat, and /v1/embeddings for vector generation.

How do I switch from GPT-4 to a local model?

You simply update the client base_url to point to your local Ollama host and change the 'model' parameter in your code to a local model name like 'llama3.2'.

Can I use this with LangChain and LlamaIndex?

Yes, this is one of the primary benefits. By setting the base_url to your local Ollama instance, you can use the standard OpenAI classes provided by these frameworks.

OpenAI Compatibility for Ollama

Name: OpenAI Compatibility for Ollama
Author: atrawog

byatrawog

0•

Data Science & ML

Integrates Ollama with the OpenAI Python library to enable seamless migration and compatibility with OpenAI-based tools.

This skill provides a standardized OpenAI-compatible interface for local Ollama instances, allowing developers to use the official OpenAI SDK, LangChain, and LlamaIndex without rewriting core application logic. By mapping Ollama endpoints to the standard /v1/* format, it facilitates an easy transition between proprietary and local models while maintaining access to familiar implementation patterns for completions, chat, and embeddings.

Key Features

010 GitHub stars

02Full support for chat completions and text generation endpoints

03Seamless integration with LangChain and LlamaIndex frameworks

04Simplified migration path from cloud APIs to local LLM infrastructure

05Local embedding generation using OpenAI-standard request formats

06Drop-in OpenAI SDK replacement for local Ollama models

Use Cases

01Prototyping AI workflows locally using the familiar OpenAI Python library syntax

02Migrating an existing OpenAI-powered application to a local Ollama backend for privacy or cost savings

03Using popular AI frameworks like LangChain or LlamaIndex with local models without writing custom adapters

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add atrawog/bazzite-ai-plugins openai

For use in Claude.ai and ChatGPT

Download Skill