GPT Wire Quest

Tracks

1. 01🧱OpenAI API Fundamentals

   0/8 lessons

   Roles, models, tokens, auth, rate limits, parameters

   Cleverness 보다 contract 를 먼저 wire. Role 과 message shape, 2026 모델 라인업, token economics, authentication, rate limit, 그리고 실제로 동작을 바꾸는 parameter 들.

   Lesson list (8)

   • 01
     Chat Completions API~22 min · chat-completions, messages, roles
   • 02
     2026 Model Lineup~22 min · models, gpt-5, o-series
   • 03
     Token Economics~22 min · tokens, cost, tiktoken
   • 04
     Responses API — 새 front door~22 min · responses-api, stateful, input
   • 05
     Authentication 과 키 관리~22 min · auth, api-keys, projects
   • 06
     Rate Limits 와 backoff~22 min · rate-limits, tpm, rpm, backoff
   • 07
     API Versioning 과 모델 pinning~22 min · versioning, deprecation, model-pins
   • 08
     Parameter Deep Dive — temperature, seed, reasoning_effort~22 min · temperature, top_p, seed, params

   ✓퀴즈 · 6 문제→

2. 02🐍Python SDK

   0/7 lessons

   Sync, async, streaming, errors, structured outputs

   Official openai Python 패키지 처음부터 끝까지. 설치, 설정, sync vs async client, response object, streaming chunk, retry, 그리고 Pydantic 으로 shape 박힌 structured output.

   Lesson list (7)

   • 01
     설치와 셋업~22 min · install, setup, venv
   • 02
     Client 초기화 — 1 process 당 1 client~22 min · client, async, httpx
   • 03
     Chat Completions — sync 호출과 multi-turn~22 min · chat-completions, sync, messages
   • 04
     Response 구조 — choices, usage, tool_calls~22 min · response, choices, usage
   • 05
     Streaming — token 단위 delivery~22 min · streaming, delta, async-iter
   • 06
     Error Handling — retry 룰~22 min · errors, retries, exceptions
   • 07
     Structured Outputs 와 Batch API~22 min · structured-outputs, pydantic, batch

   ✓퀴즈 · 5 문제→

3. 03🌊Responses API

   0/7 lessons

   Stateful session, built-in tools, semantic events

   OpenAI 가 권장하는 새 shape 의 API. 왜 존재하는지, chat.completions 와 어떻게 다른지, input format, hosted tool, custom function tool, previous_response_id 로 multi-turn session, background mode, reasoning 컨트롤.

   Lesson list (7)

   • 01
     왜 Responses API?~22 min · responses-api, rationale
   • 02
     responses.create vs chat.completions.create~22 min · responses-api, shape-diff
   • 03
     Input Format — string vs message list~22 min · input, messages, multimodal
   • 04
     Built-in Tools — web_search, code_interpreter, file_search~22 min · hosted-tools, web-search, code-interpreter, file-search
   • 05
     Custom Function Tools — schema 와 shape~22 min · function-tools, tool-loop
   • 06
     Stateful Sessions — previous_response_id 트리~22 min · sessions, previous_response_id, stateful
   • 07
     Background Mode 와 Reasoning Effort~22 min · background, reasoning, o-series

   ✓퀴즈 · 5 문제→

4. 04⏱️Streaming Deep Dive

   0/6 lessons

   SSE wire format, semantic events, token renderer

   Wire 를 가리지 말고 그대로 real-time delivery. UX 에서 streaming 이 왜 중요한지, Chat Completions 의 delta shape, Responses 의 semantic event shape, function-call streaming, raw SSE 파싱, 그리고 production-quality token renderer.

   Lesson list (6)

   • 01
     왜 Streaming 이 중요한가~22 min · streaming, ux, ttfb
   • 02
     Chat Completions Streaming — delta 라이프사이클~22 min · streaming, chat-completions, delta
   • 03
     Responses API Streaming — typed semantic events~22 min · streaming, responses-api, semantic-events
   • 04
     Function Call Streaming — fragment 조립~22 min · streaming, function-calls
   • 05
     SSE Wire Format — 100 줄짜리 spec~22 min · sse, wire-format, raw
   • 06
     Token Renderer 만들기 — SSE 와 UI 사이의 contract~22 min · renderer, ui, production

   ✓퀴즈 · 5 문제→

5. 05🔧Function Calling / Tool Use

   0/7 lessons

   Schema, strict mode, tool loop, parallel calls

   Tool 은 모델이 runtime 으로 넘어오는 통로야. Schema design, strict mode, tool decision boundary, multi-turn tool loop, parallel call, error handling, 그리고 동작을 실제로 steering 하는 description.

   Lesson list (7)

   • 01
     Tool Definition Schema — JSON Schema 위 contract~22 min · tools, schema, function-calling
   • 02
     Strict Mode — schema-enforced 호출~22 min · strict, schema-validation
   • 03
     GPT 가 tool 호출 결정하는 법 — tool_choice~22 min · tool-decision, tool_choice
   • 04
     The Tool Loop — 3 phase per turn~22 min · tool-loop, multi-turn
   • 05
     Parallel Function Calling — 한 turn 에 여러 호출~22 min · parallel-calls, concurrency
   • 06
     Tool Error Handling — 모델이 recover 하게~22 min · tool-errors, recovery
   • 07
     효과적인 Tool Description 설계~22 min · tool-descriptions, prompting

   ✓퀴즈 · 6 문제→

6. 06👁️Vision & Multimodal

   0/5 lessons

   Image in, image out, audio, PDFs

   Production multimodal — URL/base64 로 image 보내기, detail level 과 cost, gpt-image-1.5 로 image gen, audio API (TTS / 전사 / realtime), 그리고 PDF understanding 위 structured output.

   Lesson list (5)

   • 01
     Image 보내기 — URL vs base64~22 min · vision, images, input
   • 02
     Image Detail Level — cost knob~22 min · vision, detail, cost
   • 03
     Image Generation — gpt-image-1.5~22 min · image-gen, gpt-image-1.5
   • 04
     Audio APIs — TTS, 전사, Realtime~22 min · audio, tts, transcription
   • 05
     PDF Understanding + Structured Output~22 min · structured-outputs, pdf, input_file

   ✓퀴즈 · 5 문제→

7. 07⚡Raw HTTP — Building Without the SDK

   0/7 lessons

   httpx, OAuth, SSE 파싱, 커스텀 adapter

   SDK 가 길을 막을 때 — httpx 요청, request body 직접 작성, httpx.stream 으로 streaming, OAuth bearer token, SDK 없이 multi-turn tool loop, 그리고 한 앱이 여러 provider 와 대화하는 Adapter 패턴.

   Lesson list (7)

   • 01
     왜 raw 로 가는가~22 min · raw-http, rationale, sdk-limits
   • 02
     httpx Basics — Client lifecycle 와 timeouts~22 min · httpx, client, timeouts
   • 03
     Request 직접 만들기 — wire 의 contract~22 min · request-body, headers, auth
   • 04
     httpx Streaming — async iter_lines + SSE parse~22 min · httpx-stream, sse-parse
   • 05
     OAuth Authentication — non-OpenAI host~22 min · oauth, bearer-tokens, non-openai
   • 06
     SDK 없이 Multi-Turn Tool Loop~22 min · raw-tool-loop, no-sdk
   • 07
     자체 Adapter 만들기 — narrow boundary~22 min · adapter-pattern, abstraction

   ✓퀴즈 · 5 문제→

8. 08🏗️Production Agent Patterns

   0/7 lessons

   Adapters, sessions, fallbacks, cost, logging, security

   Production-grade agent wiring — provider adapter, tool-loop orchestration, session state, fallback chain, cost/rate-limit 관리, JSONL logging + replay-based 테스팅, 그리고 모든 public deploy 에 필요한 security boundary.

   Lesson list (7)

   • 01
     Adapter Pattern — wire 에, prompt 에 X~22 min · adapter, production
   • 02
     Tool Loop Orchestration — 두 layer, 두 책임~22 min · orchestration, tool-loop
   • 03
     Session Management — JSONL ground truth~22 min · sessions, state, persistence
   • 04
     Fallback Chains — 다른 failure mode 다른 path~22 min · fallback, reliability
   • 05
     Cost & Rate Limit Management — per-tenant ledger~22 min · cost, rate-limits, budgets
   • 06
     JSONL Logging & Replay-based Testing~22 min · jsonl, logging, replay-testing
   • 07
     Security & Deployment — 모델은 boundary 아님~22 min · security, deployment, prod

   ✓퀴즈 · 6 문제→