Codex and GitHub Copilot overlap: GitHub offers Codex as a third-party coding agent and as a sign-in option for the Codex VS Code extension. Compare the exact access path, workflow, execution boundary, usage billing, and private-code policy before choosing. This guidance is based on company pages linked from the profiles below.
You want one coding agent across local repository work, cloud delegation, terminal, editor, and code review workflows; the CLI can also use local OSS providers through Ollama or LM Studio.
Look elsewhere if: You need managed cloud agents to run on infrastructure you control, need a fixed task allowance, or want cloud tasks while authenticating only with an API key.
Development already runs through GitHub and supported editors, with review and agents in that workflow.
Look elsewhere if: You need the cloud agent for a repository not hosted on GitHub, or require every Copilot surface to work in an air-gapped environment.
Free and Go include limited trial access. Plus is $20 per month. Pro starts at $100 per month, with current tiers offering 5× or 20× Plus usage. Business is $20 per user monthly when billed annually or $25 month to month, with at least two users. Enterprise and Edu use sales-led pricing. OpenAI's newer rate card says Plus, Pro, and Business accounts plus most Enterprise-family accounts use token-based rates, while its Codex pricing page still says only Business and new Enterprise accounts have migrated; check the Usage panel for the meter that applies to your account. Some Plus and Pro users can add credits, while others may need to upgrade or wait for a reset. All users can run extra local tasks with an API key at standard API rates. Credit use depends on the model, input and output token mix, speed, and any additional agents. OpenAI publishes a cross-plan planning estimate of roughly $100–$200 per developer per month, with large variance; it does not define an included allowance or predict an individual bill.
Free plan. Pro is listed at $10 monthly, Pro+ at $39, and Max at $100. For organizations, Business is $19 per user monthly and Enterprise is $39 per user monthly; Enterprise requires GitHub Enterprise Cloud.
Ways to use it
Web, macOS, Windows, Linux, iOS, Terminal, VS Code
GitHub, GitHub Mobile, macOS, Windows, Linux, VS Code, Visual Studio, JetBrains, Xcode, CLI, More editors
Access
Available through Codex in ChatGPT, the desktop app, CLI, IDE extension, web, and iOS, subject to plan, platform, and workspace settings. The Codex IDE extension also supports Cursor and Windsurf; OpenAI documents separate integrations for Xcode and JetBrains IDEs. Local CLI, IDE, and SDK use can instead authenticate with an API key; API-key access does not include cloud features such as GitHub code review or Slack.
Hosted service available on GitHub, GitHub Mobile, supported editors, and the command line, plus a desktop app for macOS, Windows, and Linux.
Where it runs
Local workflows run commands on the user's device under configurable sandbox, approval, and network policies; the CLI can also use OSS models served by Ollama or LM Studio. Cloud tasks run in isolated OpenAI-managed environments; setup scripts have internet access, while agent internet access is off by default unless configured.
GitHub-hosted by default with organization and enterprise controls; supported clients can instead use third-party or local models through local BYOK without a Copilot subscription, though provider costs may apply.
Are you choosing products, or a Codex access path inside Copilot?
Standalone Codex is OpenAI's coding agent across its desktop app, CLI, IDE extension, web, iOS, and cloud tasks. GitHub Copilot is a broader GitHub, editor, command-line, and desktop-app suite with its own completion, chat, review, and agent features. GitHub exposes Codex in two distinct ways. Its public-preview third-party Codex coding agent can take asynchronous GitHub tasks and open pull requests; GitHub's agent documentation says this is available on all paid Copilot plans, but the Copilot plan page's feature table marks third-party delegation unavailable on Pro while its Pro card says Claude and Codex are included. Separately, GitHub says signing into the Codex VS Code extension with Copilot requires Pro+ or Max. The cited GitHub pages cover those two Copilot routes; use the standalone Codex sources to compare other clients and features. Choose the exact route before comparing plans.
Use standalone Codex when its app, CLI, IDE extension, or optional handoff from a local IDE session to an OpenAI cloud task should carry the work. Use Copilot's native tools when inline completion, chat, agents, and review should stay across supported editors, GitHub, its CLI, or the Copilot desktop app. If the work begins as a GitHub issue or pull request and should run asynchronously, compare Copilot's own cloud agent with the third-party Codex agent there; both can work through a pull request, but they are separate agents. Test the exact client and repository flow rather than choosing from the product names alone.
Which execution and review boundary fits the repository?
For local work, Codex separates a platform-enforced sandbox from approval policy. Its cloud tasks run in isolated OpenAI-managed environments where setup scripts have internet access and agent internet access is off by default. The Copilot desktop app can run parallel sessions in dedicated worktrees and branches, offers Interactive, Plan, and Autopilot modes, and can instead use a public-preview GitHub cloud sandbox. Copilot's separate cloud agent works in an ephemeral GitHub Actions environment, changes one selected repository and branch, and opens a pull request that must be reviewed and merged by a person. GitHub describes Codex on GitHub as a third-party coding agent using that asynchronous issue-to-pull-request workflow, with GitHub validation and policy controls. Compare the exact local, OpenAI-cloud, Copilot-app, or GitHub-agent path; the product names do not define one shared security boundary.
Does private code need a local-model or air-gapped request path?
Codex CLI documents a local OSS-model path through Ollama or LM Studio. OpenAI documents that local-inference route for the CLI. GitHub documents a separate local BYOK path for VS Code, JetBrains, Xcode, Copilot CLI, the Copilot app, and the Copilot SDK. It can use local or externally hosted models; GitHub says keys remain client-side and the route removes the Copilot API dependency, making it suitable for air-gapped environments or use without a Copilot subscription. GitHub's cited BYOK page lists those clients; choose only from that documented scope for an isolation requirement. Business or Enterprise administrators can disable local BYOK in IDEs. GitHub's separate Enterprise BYOK path is server-side and still requires a Copilot license and internet access.
Use the current price row above for subscription costs. ChatGPT Plus, which includes Codex, lists $20 monthly; Copilot Pro lists $10 monthly. Both publish variable token or credit metering rather than a repository-task allowance. OpenAI's newer rate card says Plus, Pro, and Business accounts plus most Enterprise-family accounts use token-based Codex rates, while its Codex pricing page still says only Business and new Enterprise accounts have migrated; check the account's Usage panel. Some Codex users can add credits. Copilot uses token-based AI Credits, while cloud-agent and third-party-agent work can also consume GitHub Actions minutes. Its official pages conflict on whether Pro includes third-party Codex, and signing into the Codex extension with Copilot requires Pro+ or Max. For a hard subscription-price ceiling, leave Codex credit auto-reload off and do not add credits or API billing. In GitHub's Budgets and alerts, create an AI-credit budget—bundled, or the applicable route-specific AI-credit SKU such as Copilot cloud agent—and a separate Actions budget before use; enable ‘Stop usage when budget limit is reached’ where available. A budget without that option only alerts, and GitHub warns that usage before a budget is created is not counted during the first billing cycle. Then test a representative week in the exact route and record reported usage instead of converting different units into an unsupported task count.
How may prompts and private code context be used for model training?
For Plus and Pro, OpenAI may use Codex content to improve models. ChatGPT conversations and Codex tasks can be opted out through general data controls, while full-environment Codex training has a separate control in Codex Settings that the general controls do not change. Business, Enterprise, Edu, and API inputs and outputs are not used for training by default. GitHub says personal Copilot interactions—including code context from a private repository while Copilot is active—may be used for training unless the user opts out; private repositories at rest and Business or Enterprise customer data are excluded. For OpenAI models served through Copilot, GitHub reports a zero-data-retention agreement with OpenAI. The checked public-preview page for signing into the Codex VS Code extension with Copilot does not say whether that extension path uses the same hosting arrangement. Confirm that preview route with GitHub before opening private code when the distinction matters.
