OpenAI Assistants API
OpenAI Assistants API was a programmatic interface, in beta from 2023, that let developers create persistent, stateful AI assistants with capabilities including code execution, file processing, and function calling through managed conversational threads and tool integration. OpenAI announced its deprecation in 2025 and shut it down on 26 August 2026; calls to the /v1/assistants, /v1/threads and /v1/threads/runs endpoints return errors after that date. Its replacement is the Responses API together with the Conversations API, in which assistants map to prompts, threads to conversations, and runs to responses. The API provided infrastructure for building AI assistants that maintained context across multiple interactions, used custom instructions, and accessed external tools through standardized endpoints without a separate state management implementation. The API included built-in tools: Code Interpreter for data analysis and visualization, File Search over vector stores for document retrieval — introduced with the v2 revision in April 2024 in place of the earlier Retrieval tool — and function calling for custom API interactions that extended assistant capabilities beyond text generation. It managed conversation threads, message history, and run states, so that context preservation and session handling did not have to be implemented by the caller. Further features included file uploads, assistant customization through instructions and model selection, and parallel tool execution for multi-step tasks. Run execution was asynchronous: the caller either polled the run status or consumed a streamed response. Threads and runs were not covered by OpenAI's webhook events. The managed service removed part of the infrastructure and state management work from conversational AI development, which is why it ended up in production systems; those systems had to be rebuilt on the Responses API before the shutdown date, and OpenAI provided no automated migration of threads to conversations.
Related terms
Related services: Agentic AI consulting.
Ready to put agentic AI to work?
Book a free 45-minute consultation. We'll map one real process worth automating with production-grade AI.