Documentation - OpenAI API

OpenAssistantGPT is built on OpenAI's Responses API. Every agent you create runs on it directly, using your own OpenAI API key.

Migrating from the Assistant API

OpenAI is retiring the older Assistant API that traditional chatbots were built on. If you still have a classic chatbot, open it from your dashboard, choose Convert to Agent from the three-dot menu, and it will move over to the Responses API.

Converting keeps the same chatbot: the id does not change, so your embed code, links, conversation history and API integrations all keep working. There is nothing to update on your website.

Two things behave slightly differently after converting:

• Web search switches from our own search provider to OpenAI's built-in web search, which is billed through your OpenAI account rather than against web search credits.
• Knowledge base files are served through an OpenAI vector store. If a chatbot's files cannot be located, conversion is blocked rather than silently producing an agent that has forgotten everything it knew.

Models

You can use any model your OpenAI account has access to that supports the Responses API. Newer accounts sometimes cannot select GPT-4 class models until they reach usage tier 1 — see OpenAI's usage tiers for details.

Tools

Agents can use file search over your uploaded files, the code interpreter for data and chart generation, web search, and any custom GET or POST actions you attach. All of this is configured from the agent's settings page and stored by OpenAssistantGPT, so it stays intact regardless of what happens to your OpenAI assistant objects.