OpenAI Codex
Also known as: Codex CLI
OpenAI's agentic coding system across desktop, CLI, IDE, web, cloud and remote, with an OS-enforced sandbox and independent approval policy, Auto-review, skills and plugins, browser and computer use, and OpenTelemetry audit export.
Codex is OpenAI's agentic coding system, an umbrella for a family of surfaces that share a single account context. It is designed not to autocomplete lines but to act as an agent. It reads a codebase, plans an approach, edits files across the project, runs shell commands and tests, and proposes finished pull requests, sustaining long autonomous sessions across many sequential tool calls.
Developers can reach Codex wherever they work. The Codex CLI is a terminal-native agent, open source and built in Rust, that runs locally, edits and runs code in a chosen directory, and ships with structured plan, execute, and review commands plus sandboxed execution and approval modes you control.
An IDE extension brings the same agent into editors like VS Code, Cursor, and Windsurf, and a cloud agent, reachable through ChatGPT, runs tasks asynchronously in isolated sandboxed containers preloaded with your repository, so many tasks proceed in parallel.
The ChatGPT desktop app, on macOS, Windows and Linux, acts as a command center for running agent threads side by side, and the ChatGPT mobile app and a remote mode let you kick off and approve work from a phone.
Around the core loop, Codex adds higher level workflows. Skills extend it beyond code into tasks like code understanding, prototyping, and documentation aligned to a team's standards. Automations let it pick up routine work such as issue triage, alert monitoring, and CI without being asked, and a separate review agent can check changes before they are committed. An AGENTS.md file in a repository tells Codex how to navigate the code, run tests, and follow project conventions, much like a README for the agent.
Codex is powered by OpenAI's frontier GPT models, trained with an agentic focus so the model handles multi-step tool use and self-checks its work before submitting, and local clients can also be pointed at other providers' models, such as Mistral's API or a local Ollama endpoint. It supports the Model Context Protocol for connecting external tools, including parallel tool calls, and runs work inside sandboxed environments while letting users verify output through citations, logs, and test results. It is included with ChatGPT plans from Free and Go upward rather than sold as a separate product, and can also be used through an API key billed per token.
From our comparison video
Google Jules vs OpenAI Codex (2026): Which Should You Choose?
Our 5 minute video puts OpenAI Codex side by side with Google Jules. It covers what each one does and how each is priced, then when to choose which.
Recorded in October 2026. The profile here is live and updates as research is verified.
Vendor details
Canonical URL
https://learn.chatgpt.com/codex
Category
Coding agent
Company status
first party product
Use cases & customers
Target customers
Deployment options
Integrations
MCP client support with enterprise approved server lists, plus a Codex MCP Server, App Server and SDK for building on the harness. Third party integrations cover GitHub, GitLab in beta, Slack and Linear, and a GitHub Action and non-interactive mode serve CI. Hooks fire on lifecycle events, Site tools work via WebMCP, and plugins and connectors require approval for destructive tool calls. The agent runs shell commands under an OS enforced sandbox, uses a web search tool with cached, indexed, live or disabled modes, and has Browser and Computer Use capabilities plus a browser extension.
In practice
You want to hand off a refactor and keep coding instead of babysitting it. Codex's cloud agent runs the task asynchronously in an isolated sandbox preloaded with your repo, then proposes a pull request for review.
You live in the terminal and don't want a browser or IDE in the loop. The open-source Codex CLI runs the agent locally, editing files and running tests in your chosen directory with approval modes you control.
You want routine work handled without prompting it each time. Codex Automations pick up issue triage, alert monitoring, and CI tasks on their own, while a separate review agent checks changes before they're committed.
Sources & related URLs
Related / legacy domains
Agentic Index coverage score
13.0 / 14 capabilities · 93%
| Integrations & Tool Calling | Full |
|---|---|
|
MCP connects outside services, with approved server lists under enterprise control, and plugins and connectors extend the agent with tool calls that have side effects, which require approval when they advertise destructive annotations. Third party integrations cover GitHub, GitLab, Slack and Linear. The agent runs shell commands under an OS sandbox, uses a web search tool with cached, live and disabled modes, and reaches the browser and computer use surfaces. Sourcelearn.chatgpt.com/codex/extend/mcp, /codex/third-party and /codex/agent-approvals-securityread 2026-08-30 |
|
| Workflow Orchestration | Full |
|
The agent sustains long autonomous sessions across many sequential tool calls, with configurable subagents, git worktrees for parallel work, projects and chats organizing multi step work, long running work and scheduled tasks, and the desktop app running agent threads side by side. The cloud agent works asynchronously in isolated containers so many tasks proceed in parallel, and multi agent tooling is among the stable CLI surfaces. Sourcelearn.chatgpt.com/codex/agent-configuration/subagents, /codex/environments/git-worktrees and /codex/long-running-workread 2026-08-30 |
|
| Knowledge Grounding & RAG | Partial |
|
Context is assembled for each run. The agent reads the repository, resolves @ mentioned files, calls MCP servers for outside sources and uses a web search tool whose cached mode draws on a web index OpenAI maintains. AGENTS.md and Rules give the agent standing instructions. OpenAI does not describe an index that Codex keeps over the customer's own code or documents. Sourcelearn.chatgpt.com/docs/agent-configuration/agents-md, /docs/extend/mcp and the docs llms.txt indexread 2026-09-29 |
|
| Human Oversight & Guardrails | Full |
|
Two independent layers govern the agent: sandbox mode controlling what it can technically do, and approval policy controlling when it must ask, with presets from read-only through workspace-write to danger-full-access and a granular policy that keeps chosen approval categories interactive while auto rejecting others. An Auto-review reviewer agent can evaluate eligible approval requests against a published policy checking for data exfiltration, credential probing, security weakening and destructive actions, failing closed on parse or session errors. Destructive MCP and app tool calls always require approval, .git, .agents and .codex are protected read-only inside writable roots, and enterprise managed requirements override local settings. Sourcelearn.chatgpt.com/codex/agent-approvals-security and /codex/sandboxing/auto-reviewread 2026-08-30 |
|
| Security, Identity & Governance | Full |
|
OpenAI's trust portal at trust.openai.com lists SOC 2 Type 2, SOC 3, ISO/IEC 27001, 27017, 27018, 27701 and 42001, CSA STAR and FedRAMP 20x, with the SOC 2 report scoped to the API Platform, ChatGPT Enterprise, ChatGPT Edu and ChatGPT Team, the plans and API key path Codex runs on, though the portal does not name Codex separately. Access controls include managed configuration that overrides local settings, roles and workspace permissions, SCIM and EKM on Enterprise, workload identity federation, service accounts, plugin, connector and skill controls, approved MCP server lists, a Compliance API with audit events, Private Link, IP allowlisting and mutual TLS. Sourcetrust.openai.com and learn.chatgpt.com/docs/enterprise/managed-configurationread 2026-09-29 |
|
| Observability & Auditability | Full |
|
Opt in OpenTelemetry export emits an event catalog that includes codex.tool_decision, recording approved or denied and whether the source was configuration or the user, codex.tool_result with duration, success and output snippet, codex.api_request, codex.sse_event and codex.user_prompt, with counter and duration histogram metrics, exportable over OTLP to a collector the customer controls. Enterprise adds a Compliance API and audit events, workspace analytics and an Analytics API. Telemetry is off by default and prompt text is redacted unless explicitly enabled. Sourcelearn.chatgpt.com/codex/agent-approvals-security and /codex/enterprise/compliance-apiread 2026-08-30 |
|
| Memory & State Persistence | Full |
|
Codex writes its own memories: once enabled, it turns useful context from eligible prior chats into local memory files under CODEX_HOME, redacting secrets and updating in the background after a chat goes idle. Memories live in the local Codex home, /memories chooses per chat whether a chat may read existing memories or feed future ones, and managed configuration can pin the feature on or off. Unused memories retire from consolidation after memories.max_unused_days, 30 by default and capped at 365, memories.max_rollout_age_days bounds the chats considered, and the files can be opened for review. The feature is off by default. AGENTS.md holds standing rules, separate from memory. Sourcelearn.chatgpt.com/docs/customization/memories and /docs/config-file/config-referenceread 2026-09-29 |
|
| Deployment & Data Residency | Full |
|
Local environments run entirely on the developer's machine through the CLI, IDE extension and desktop app under an OS enforced sandbox using Seatbelt on macOS, bwrap and seccomp on Linux and a native Windows sandbox or WSL2, with dev container support and a published secure devcontainer reference; cloud environments run in isolated OpenAI managed containers with configurable internet access and domain allow lists. Enterprise adds Private Link, IP allowlisting, mutual TLS, IP egress ranges and Amazon Bedrock as a model provider. Sourcelearn.chatgpt.com/codex/environments/modes, /codex/agent-approvals-security and /codex/amazon-bedrockread 2026-08-30 |
|
| Prebuilt Agents, Templates & Packs | Full |
|
Skills and plugins extend Codex, with dedicated build guides, a plugin architecture and submission path, enterprise plugin management, and plugin and skill controls for administering approved catalogs, and a Claude Code plugin can be submitted as well. Subagents are configurable per project, and Codex Micro ships as a packaged capability. Sourcelearn.chatgpt.com/codex/skills-and-plugins, /codex/build-plugins and /codex/enterprise/plugin-managementread 2026-08-30 |
|
| Triggers & Channel Coverage | Full |
|
Codex runs in the ChatGPT desktop app, a Codex CLI, an IDE extension, ChatGPT on the web, Codex cloud, and a Remote mode driven from a phone. Scheduled tasks run work on a timetable, long running work is supported, a GitHub Action and non-interactive mode cover CI, and third party integrations reach GitHub, GitLab in beta, Slack and Linear, with notifications closing the loop. Sourcelearn.chatgpt.com/codex/automations, /codex/github-action and /codex/third-partyread 2026-08-30 |
|
| Model Flexibility & Routing | Full |
|
Local Codex clients let the customer choose the model maker, accepting custom model providers in config.toml, each defined by base URL, wire API and authentication, and OpenAI's own example configures Mistral's API and a local Ollama endpoint beside an OpenAI proxy. OSS mode runs Codex against a local open source provider, Ollama or LM Studio, chosen per run with --local-provider or set as the default. Within OpenAI's own family the model is selectable per session and per workspace. Amazon Bedrock is a built in provider that serves OpenAI models on AWS. Sourcelearn.chatgpt.com/docs/config-file/config-advanced (Custom model providers, OSS mode) and /docs/modelsread 2026-09-29 |
|
| APIs, SDKs & MCP Extensibility | Full |
|
Developers build on a Codex SDK, an App Server, an MCP Server, a GitHub Action and a non-interactive mode, alongside MCP client support with approved server lists, hooks that fire on lifecycle events, Site tools via WebMCP, Record and Replay, and a Python SDK with auth. The CLI is open source in Rust at github.com/openai/codex with a published default reviewer policy, and the documentation offers an llms.txt index and a markdown version of every page. Sourcelearn.chatgpt.com/codex/codex-sdk, /codex/mcp-server and /codex/open-sourceread 2026-08-30 |
|
| Testing, Debugging & Optimization | Partial |
|
Codex reviews code changes before commit, a Codex Security plugin and CLI run scans, deep scans and bulk scans in CI with a security workbench and finding export, and the agent runs tests and iterates on failures; Auto-review evaluates individual agent actions against a risk policy at runtime. Those tools check the customer's code and gate the agent's actions, and there is no harness to evaluate, replay or regression test the agent's own behavior. Sourcelearn.chatgpt.com/codex/code-review and /codex/securityread 2026-08-30 |
|
| Browser & Computer Use | Full |
|
Browser and Computer Use let the agent work in a browser and operate a computer, alongside a browser extension for signed in browser workflows and Appshots. Browser and Computer Use activity is treated as its own traffic surface outside the command network proxy, with its own feature settings and workspace policies, and managed configuration lets administrators set feature requirements for browsers and Computer Use. Sourcelearn.chatgpt.com/codex/computer-use, /codex/browser and /codex/agent-approvals-securityread 2026-08-30 |
|
The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded
Product updates
Codex CLI 0.169.0 makes non interactive runs require explicit project trust: codex exec keeps project configuration and MCP servers disabled in untrusted directories, even with --sandbox workspace-write. Users pass --trust to codex exec or codex review to save trust for a project on purpose. The release also extends the opt in Cloud connection on macOS and Linux, so users can send images and choose a model and reasoning effort.
Bears on: Human approval / guardrails
View sourceCodex Security 0.10.0, the CLI and SDK for the Codex security scanning agent, adds codex-security scans resume SCAN_ID to continue an interrupted scan from a saved checkpoint with its usage and budget intact, and a --max-total-cost option that shares one estimated budget across repositories or components. It also imports saved ArmorCode SAST findings into the local workbench and lets SDK callers publish selected workbench findings to Cloud; MCP registration now needs --all, --agent NAME or --package NAME.
Bears on: Agent capability
View sourceCodex CLI 0.164.0 lets enterprise MCP connections authenticate with eligible Codex account grants or a dedicated sign in, including OAuth clients that require secrets. The same release masks credentials during supported remote Bash commands and Zsh snapshot sessions, including credentials from custom providers.
Bears on: Security / enterprise
View sourcePricing
Included in ChatGPT plans. Free is $0 and Go is $8 a month, with GPT-6 Luna in the desktop app subject to rollout. Plus is $20, Pro starts at $100 ($100, $200 or $500), and Business is $20 per user a month billed annually ($25 monthly). Enterprise and Edu are sold through sales, and an API key pays per token.
A ChatGPT plan subscription with an included Codex usage quota, plus usage beyond the quota paid with ChatGPT credits or, on the API key path, at per token API rates.
Included quota
Included Codex usage scales with the ChatGPT plan tier (Plus, Pro or Business). Exact task and credit allowances vary by plan and by OpenAI's rate card.
What is public
Every plan is published. Free is $0 and Go $8 a month, with GPT-6 Luna in the desktop app subject to rollout. Plus at $20 adds Codex on the web, CLI, IDE extension and iOS, with GPT-6.1 Sol and GPT-6 Luna. Pro starts at $100 a month in $100, $200 and $500 tiers, and the $500 tier adds Ultrafast mode on GPT-6.1 Sol. Business is $20 per user a month billed annually for two or more users, or $25 monthly. Enterprise and Edu are sold through sales with SCIM and EKM, and an API key path bills at API pricing. Plus and Business extend usage with ChatGPT credits, and ChatGPT Work shares the same pricing, credits and limits as Codex.
Billing mechanics
A bundled subscription plus usage. A ChatGPT plan includes a quota of Codex usage. Beyond it, usage is metered against the plan's credits or OpenAI's API rate card. The Codex CLI is open source and can run against the API with the customer's own key.
Cost watchouts
Agent heavy or long running tasks can drive metered usage well above the base ChatGPT plan cost.
Variable cost rationale
Codex work is metered against the ChatGPT plan. Each plan includes a usage allowance, and work beyond it draws ChatGPT credits or, on the API key path, bills at per token API rates. The bill grows with how much agent work developers run, and long running or agent heavy tasks use the most. The plan's allowance is the main limit, and Pro's $100, $200 and $500 tiers carry different allowances. ChatGPT Work follows the same pricing, credits and limits as Codex. No hard cap on credit or API spend is published.
Additional watchouts
There is no per seat Codex price. The cost is the ChatGPT plan plus whatever credits or API usage the agent workloads run up.
Overage / add-ons
Usage beyond the plan's allowance is metered, through ChatGPT credits on the plan or pay as you go API billing.
Sales call required
No, self serve available
Free / trial
ChatGPT Free includes Codex in the desktop app on GPT-6 Luna at Standard speed, subject to rollout
Lowest paid plan
ChatGPT Go at $8 per month
Commercial notes
There is no standalone Codex plan. Access comes with ChatGPT Free, Go, Plus, Pro, Business, Enterprise or Edu, or through an API key billed per token. Pro comes in three tiers from $100.
Key ambiguities
Codex's included quota and credit behavior differ by ChatGPT plan and follow OpenAI's rate card, so no single entry price represents the cost.
Cancellation / refund
Billed through the ChatGPT subscription under OpenAI's standard subscription terms.
Missing data
Exact Codex quotas per plan are not published as one figure, and metered rates are spread across the credit and API rate cards.
Related vendors
- Cognition — Maker of Devin, an autonomous AI software engineer
- 10Web — Agentic website platform whose AI agents build, host and manage…
- AgentUI — Managed AI app builder for operations teams that generates and hosts…
- Aider — Open source, model agnostic terminal coding agent that edits your…
- Amp — Amp is a coding agent from Amp Frontier Corporation that runs each…
- Anthropic Claude Code — Anthropic's agentic coding system across terminal, desktop, IDE, web…
Alternatives to OpenAI Codex
The closest documented capability profiles to OpenAI Codex among coding agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.
- Base4413.0 / 14Matches OpenAI Codex across all 14 documented capabilities
- Capy12.5 / 14A lighter documented profile than OpenAI Codex
- Cognition13.5 / 14Fuller documented coverage on Knowledge Grounding & RAGOpenAI Codex vs Cognition →
- Cursor12.5 / 14A lighter documented profile than OpenAI CodexOpenAI Codex vs Cursor →
- GitHub Copilot13.5 / 14Fuller documented coverage on Knowledge Grounding & RAGOpenAI Codex vs GitHub Copilot →
- Google Antigravity12.5 / 14A lighter documented profile than OpenAI Codex
Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded