The OpenAI API gives developers programmatic access to GPT-5.5, GPT-5.4, and the full range of OpenAI models through two primary interfaces: the Responses API (stateful, agentic, recommended for new projects) and the Chat Completions API (stateless, wide compatibility). With GPT-5.5 launching in April 2026 as the new flagship and open-weight gpt-oss models released for local deployment, the ecosystem now spans everything from sub-cent commodity inference to frontier research-grade reasoning. The core paradigm shift is agents and tools over raw completions: built-in web search, computer use, MCP server integration, and background async execution are first-class API features, not just ChatGPT-only capabilities.
What This Cheat Sheet Covers
This topic spans 22 focused tables and 214 indexed concepts, 134 flashcards. Below is a complete table-by-table outline of this topic, spanning foundational concepts through advanced details.
A jump-to index of every table row in this cheat sheet.
An interactive map of every table and concept in this topic.
Table 1: API Endpoints
The OpenAI REST API is versioned under https://api.openai.com/v1. The Responses API (/v1/responses) is the recommended path for new agentic applications; the Chat Completions endpoint remains fully supported for compatibility. All endpoints require an Authorization: Bearer <API_KEY> header.
| Endpoint | Example | Description | |
|---|---|---|---|
POST /v1/responses | • Recommended stateful agentic endpoint • supports built-in tools (web search, computer use, MCP), previous_response_id chaining, background mode, and store. | ||
POST /v1/chat/completions | • Stateless multi-turn conversation endpoint • still fully supported • use for legacy integrations or when stateless behavior is required | ||
POST /v1/embeddings | Converts text to vector representations for semantic search, clustering, and retrieval. | ||
POST /v1/images/generations | • Generates images using gpt-image-2 (flagship) or gpt-image-1.5• DALL-E 2/3 shut down May 12 2026. | ||
POST /v1/images/edits | • Inpaints or edits images using a mask • only gpt-image-2 supported | ||
POST /v1/audio/speech | • Text-to-speech using gpt-4o-mini-tts or other TTS models• returns audio stream | ||
POST /v1/audio/transcriptions | • Speech-to-text using whisper-1 or gpt-4o-transcribe• accepts audio file upload | ||
POST /v1/audio/translations | Translates audio to English text using whisper-1. | ||
wss://api.openai.com/v1/realtime | • WebSocket-based low-latency bidirectional audio/text • gpt-realtime-2 (reasoning) and gpt-realtime-translate (speech-to-speech translation). | ||
POST /v1/batches | • Async offline bulk processing at 50% discount • results within 24 h via JSONL file output | ||
POST /v1/files | Uploads files for fine-tuning datasets, assistants, or batch inputs. | ||
POST /v1/moderations | Classifies content against OpenAI's usage policies using omni-moderation-latest. | ||
POST /v1/fine_tuning/jobs | • Creates fine-tuning jobs • self-serve platform being wound down — no new customers since May 7 2026. | ||
GET /v1/models | • Lists all available models • GET /v1/models/{model} returns metadata for a specific model | ||
POST /v1/conversations | • Persistent conversation threads for Responses API • Assistants API replacement (Assistants shutdown Aug 26 2026). |