Built by Metorial, the integration platform for agentic AI.
List files uploaded to OpenAI, optionally filtered by purpose (e.g. "fine-tune", "assistants"). Returns file metadata including ID, name, size, and purpose.
List all available OpenAI models, or retrieve details about a specific model. Useful for discovering available model IDs, owners, and capabilities before making API calls.
Delete a vector store by its ID. This permanently removes the store and its indexed data.
Generate vector embeddings for text input using OpenAI embedding models (text-embedding-3-small, text-embedding-3-large). Useful for search, RAG, clustering, and semantic similarity. Supports configurable output dimensions.
Generate text using OpenAI chat completion models (GPT-5, GPT-4o, etc.). Supports multi-turn conversations, system instructions, structured JSON output, and configurable generation parameters.
Submit a batch of API requests for asynchronous processing at reduced cost. Supports Responses, chat completions, embeddings, completions, and moderations endpoints. The input must be a JSONL file uploaded via the Files API.
Generate a response using the OpenAI Responses API, the primary gateway for all model families. Supports text generation, built-in tools (web search, file search, code interpreter), function calling, structured output, and reasoning models with configurable effort levels.
Classify text against OpenAI's content policy categories. Returns flagged status and per-category scores for hate, harassment, self-harm, sexual, and violence content. Useful for filtering harmful content in user-generated input.
Retrieve metadata for a specific file by its ID. Returns file details including name, size, purpose, and status.
Retrieve the status and details of a fine-tuning job, or list all fine-tuning jobs. Includes training metrics, timestamps, and the resulting fine-tuned model name.
Cancel a batch processing job that is in progress. Already-completed requests within the batch will still be available.
Create a managed vector store for uploading, chunking, and searching files. Vector stores power file search in the Responses API and support hybrid search (semantic + keyword).
Search a vector store using natural language queries. Returns ranked results with relevance scores. Supports configurable result limits and score thresholds.
List all vector stores in your OpenAI project. Returns store metadata, file counts, and status information.
Delete a file from your OpenAI account by its ID. The file will no longer be available for use with fine-tuning, vector stores, or other features.
Create a new fine-tuning job to customize an OpenAI model on your training data. Supports supervised fine-tuning and direct preference optimization (DPO). Configure hyperparameters such as epochs, batch size, and learning rate.
Retrieve the status and details of a batch processing job, or list all batches. Returns processing progress, output file IDs, and error information.
Cancel a running fine-tuning job. The job must be in a cancellable state (e.g. "validating_files" or "running").
Generate images from text prompts using OpenAI's image generation models (e.g. DALL·E 3, gpt-image-1). Returns generated image content as Slate attachments when the API returns base64 data. Supports configurable size, quality, and style.
List the input items used to create a stored OpenAI response.
Attach an uploaded OpenAI file to a vector store so it can be indexed for file search.
Translate speech from an uploaded audio file into English text.
Retrieve parsed content chunks for a file attached to a vector store.
Delete a stored OpenAI response by ID.
Detach a file from a vector store. This removes it from the vector store index but does not delete the uploaded file.
Retrieve status and metadata for one file attached to a vector store.
Retrieve the raw content of an uploaded OpenAI file, such as JSONL batch output, fine-tuning data, or uploaded text.
List files attached to a vector store, optionally filtered by processing status.
Upload a file to OpenAI for use with batch processing, fine-tuning, vector stores, vision, or user data workflows. Accepts plain text content or base64-encoded bytes.
Retrieve a stored OpenAI response by ID, including status, model output items, and token usage.
Retrieve details for a specific OpenAI vector store, including processing status and file counts.
Generate spoken audio from text using OpenAI text-to-speech models.
Update a vector store name, expiration policy, or metadata.
Transcribe speech from an uploaded audio file into the source language text.