/Home/Explore/openai/codex

openai/codex

SchemaModels

OpenAI's Codex CLI in headless mode — structured JSON events and sandboxed execution. Serves GPT models and open-weight models like GLM, DeepSeek, and Kimi with tuned context windows.

PlaygroundAPIExamples

Open in Dashboard

Input

Form

Title

Instructions*

You are a senior software engineer. Work autonomously: plan, edit files, run tests, and report back with a concise summary.

Default Model

Claude Fable 5.1Claude Opus 5Claude Opus 4.8Claude Opus 4.7Claude Opus 4.6Claude Opus 4.5Claude Sonnet 5Claude Sonnet 4.6Claude Sonnet 4.5Claude Haiku 4.5GPT-5.6 SolGPT-5.6 TerraGPT-5.6 LunaGPT-5.5GPT-5.4GPT-5.2GPT-5GPT-5 MiniGPT-5 NanoGemini 3.5 FlashGemini 2.5 FlashGemini 2.5 ProGrok 4.6Grok 4.5Muse Spark 1.2InklingKimi K3Kimi K2.7 CodeDeepSeek V4 ProGLM-5.3GLM-5.3 FlashGLM-5.2

First Message

Clone github.com/acme/api, find the failing test, and fix it.

Auto Run

Additional settings8

Sandbox Provider

Daytona (default)E2B

Description

Entrypoint

MCP Servers

Skills

Metadata

Shared Folder

GPU Accelerator

ResetSign in to run

Result

Idle

PreviewJSON

CompletedCodex2m 14sExample run

You

Clone github.com/acme/api, find the failing test, and fix it.

Agent

Cloned the repo, reproduced the failing test in tests/test_auth.py, patched the token-expiry check, and pushed the fix — 42/42 tests passing.

Input tokens

48,211

Output tokens

6,402

Credits billed

$0.87