Llmswap

LLMSwap is an open source Python SDK, command line tool and local Arena for working with ten LLM providers through one interface. The supported providers are OpenAI, Anthropic, Gemini, Cohere, Perplexity, IBM watsonx, Groq, Ollama, xAI and Sarvam. Users bring their own provider keys and keep control of routing, and can switch models without rewriting the surrounding application. The package centers on three modes of use. Ask sends a prompt to any supported provider through the same Python API or CLI. Compare runs several models side by side in the local Arena. Best Answer, introduced in version 5.7.0, generates two to five independent responses and asks a judge model to synthesize one result, returning the answer along with the level of agreement, points of disagreement and cautions. It also supports function calling and local stdio MCP so models can reach tools and data. For operations it provides provider diagnostics, token usage, latency, optional caching, fallback and cost estimates. It suits developers who want to avoid lock in to one provider, researchers comparing model outputs, and teams that want a lightweight alternative to a hosted gateway. The package is distributed on PyPI under the MIT license and the source is on GitHub. The software is free to install and use. Costs come only from the model providers whose API keys the user supplies.