OpenAI

1. Text generation

   Generating text with the Responses API: input, instructions, model selection, prompt roles, output text, prompt engineering basics.

2. Responses API migration

   Responses API vs Chat Completions, response items, built-in tools, migrating existing integrations.

3. Conversation state

   Managing multi-turn state, previous_response_id, Conversations, manually passing history, store parameter, context window management.

4. Function calling

   Defining function tools with JSON schema, tool calls and tool outputs, parallel calls, tool_choice, strict mode.

5. Structured outputs

   JSON schema constrained output, response_format/text.format, Pydantic and Zod schemas, refusals.

6. Streaming responses

   Server-sent events streaming of response deltas, event types, streaming function call arguments.

7. Reasoning models

   o-series and GPT-5 reasoning models, reasoning effort, reasoning tokens, summaries, prompting reasoning models.

8. Built-in tools: web and file search

   Web search and file search tools, vector stores, citations/annotations, retrieval-augmented generation.

9. Embeddings

   Vector embeddings, embedding models, dimensions, similarity search, use in RAG and clustering.

10. Images and vision

    Image inputs for understanding, image generation, detail levels, image token costs.

11. Audio and voice

    Speech-to-text, text-to-speech, audio input and output in models, voice agents.

12. Realtime API

    Low-latency speech-to-speech sessions over WebRTC and WebSocket, events, turn detection.

13. Agents

    Agent building blocks, agents SDK, handoffs, guardrails, tracing, orchestrating tools.

14. Prompt caching

    Automatic prefix caching, cached input pricing, cache keys and structuring prompts for hits.

15. Evals

    Defining evaluation datasets and graders, running evals, testing prompt and model changes.

16. Fine-tuning

    Supervised, preference and reinforcement fine-tuning, training data format, evaluating fine-tuned models.

17. Batch API

    Asynchronous batch jobs, JSONL input files, 24-hour completion window, cost savings.

18. Moderation

    Content moderation endpoint, categories, safety classification of inputs and outputs.

19. Rate limits

    RPM, TPM and usage tiers, rate limit headers, 429 handling, exponential backoff.

20. Error codes

    HTTP error codes and causes (401, 403, 429, 500, 503), troubleshooting and retries.

Each section links to the official page. Titles and summaries by Extra Credit; the docs belong to their authors.

Source: endoflife.date

• LLM engineering
• MCP
• Mistral AI
• MLflow
• Ollama
• ONNX
• OpenCV
• PyTorch
• Ray
• Replicate
• scikit-learn
• spaCy