Build a provider-neutral chatbot

This minimal terminal chat keeps message history in the current process and can switch providers without changing the conversation loop.

Install and configure

python -m pip install llmswap
export OPENAI_API_KEY="your-key"

Create chatbot.py

from llmswap import LLMClient, LLMSwapError


def main():
    client = LLMClient(provider="openai", model="gpt-5.6")
    messages = []

    print("Type /quit to exit")
    while True:
        prompt = input("You: ").strip()
        if not prompt:
            continue
        if prompt == "/quit":
            break

        messages.append({"role": "user", "content": prompt})
        try:
            response = client.chat(messages, cache_bypass=True)
        except LLMSwapError as error:
            print(f"Error: {error}")
            messages.pop()
            continue

        print(f"Assistant: {response.content}")
        messages.append({"role": "assistant", "content": response.content})


if __name__ == "__main__":
    main()

Run it:

python chatbot.py

Change providers

Change only client construction:

client = LLMClient(provider="anthropic", model="claude-sonnet-5")

Or auto-detect the first configured provider:

client = LLMClient()

Production considerations

• Keep API keys in an environment or secret manager, not source code.
• Set fallback=False if data must never move to another configured provider.
• Bound conversation history according to the selected model’s context window.
• Validate model output before it triggers external actions.
• Add application-level authentication, rate limits, timeouts, and observability.
• Review provider retention and data-use terms for the account being used.

For tool calling, async access, and Best Answer, see the Python SDK guide.