Two different kinds of product
ChatGPT is a general assistant. Plus includes image generation, expanded deep research, voice with video, projects, memory and extensions for Excel, Word, PowerPoint and Google Sheets; Codex is one part of it, sharing usage with ChatGPT Work. GitHub Copilot is a coding tool: GitHub's docs scope it to code, repositories and development concepts. Image and video generation are not documented for Copilot, and voice only as dictating prompts in Copilot CLI.
Where each one works
Copilot has extensions for VS Code, Visual Studio, JetBrains IDEs, Xcode, Eclipse, Neovim, Zed, SQL Server Management Studio and Raycast, with agent mode in VS Code, Visual Studio, JetBrains, Eclipse and Xcode. GitHub says all plans include Copilot CLI and the GitHub Copilot desktop app, and it also works on github.com and GitHub Mobile. Features vary by editor.
On Plus, the Codex IDE extension signs in with your ChatGPT account and runs in VS Code, VS Code Insiders, Cursor and Windsurf; Xcode and JetBrains IDEs connect through their own integrations. Code completions as you type are not documented for Codex on any ChatGPT plan, and no page says they are absent, so they are not verified. Copilot Pro documents unlimited inline suggestions and next edit suggestions that use no AI credits.
Agents in the cloud and pull request review
Codex Cloud runs coding tasks in OpenAI-managed cloud environments, keeps going while your computer sleeps, and lets you commit or open a pull request when it is ready. On Plus it is subject to rollout and counts toward plan limits. Copilot's cloud agent researches a repo, plans, codes in a cloud development environment and opens a pull request. GitHub documents its limits: GitHub-hosted repos only, one repo, branch and pull request per task, and a 59-minute session limit.
For reviews on ChatGPT Plus, you comment @codex review on a GitHub pull request or turn on automatic reviews. Codex posts a GitHub review that flags P0 and P1 issues, with an optional Security Review in research preview. Copilot code review runs on GitHub pull requests, manually or automatically for pull requests you create, and on file diffs in editors. GitHub prices each review in credits: about $0.05 to $1 on the Lite setting, or $0.25 to $5 on Balanced. Documentation cannot show whether the code works or a review catches the bug that matters: give both the same task from the code spec kit, and open a pull request with the bug from the code repair kit to see what each review flags.
Models, limits and extra usage
ChatGPT Plus lists only OpenAI models: GPT-6.1 Sol, GPT-6 Astra, GPT-6 Sol, GPT-6 Luna, the GPT-5.6 family and legacy models. Context is 54K tokens for GPT Instant and 256K for reasoning models. OpenAI describes Codex on Plus as included, with optional credits, and does not give its limits as numbers on the pages we checked.
Copilot Pro lets you choose models from Anthropic, OpenAI, Google, xAI, Moonshot and Microsoft, among them Claude Sonnet 5.5, GPT-6 Luna, Gemini 3.8 Flash and Grok 4.7. Pro+ and Max add premium models such as Claude Opus 5.5, Claude Fable 5.1, GPT-6.1 Sol and GPT-5.6 Sol. Pro includes 1,000 base plus 500 flex AI credits a month at $0.01 a credit (GitHub says the flex part may change). Chat, agent mode, code review, the cloud agent and the CLI use credits, and extra use is billed against a dollar budget you set. An optional 1 million token context works for supported models in VS Code and Copilot CLI only, and uses more credits.
Price and billing
ChatGPT Plus is $20 a month and Copilot Pro is $10 a month. Neither offers new subscribers a yearly price: OpenAI does not support annual billing for Go, Plus or Pro, and GitHub's annual Pro plans are legacy, kept by existing subscribers until they expire. Neither documents whether US sales tax is included, so treat both as before tax. Verified students get a separate free Copilot Student plan, and verified teachers and popular open source maintainers may get Pro free.
Privacy, training and ads
Both use your data for training by default, and both let you turn it off. In ChatGPT, switch off "Improve the model for everyone" under Settings, Data controls. For Copilot, GitHub says that since April 24, 2026 it may train on interaction data from Free, Pro, Pro+ and Max; set "Allow GitHub to use my data for AI model training" to Disabled. You can request a copy of your ChatGPT data from settings or OpenAI's Privacy Portal. No Copilot chat history export is documented. Ads may appear on ChatGPT Free and Go, and Plus has none. Advertising is not mentioned on the Copilot pages we checked.
Free tiers and higher plans
Copilot Free includes 2,000 completions a month and limited chat and agent use through an AI-credit allowance, with auto model selection only (GitHub's plans FAQ still says 50 chat requests a month, so the Free limit is not verified). It has no cloud agent, no pull request review and no Memory. ChatGPT Free and Go ($8 a month) list Codex as limited, but Codex Cloud is not included, pull request review is not stated, and OpenAI's pages conflict on whether the editor extension is included.
If you outgrow the $10 or $20 plan, Copilot Pro+ ($39 a month) gives 7,000 credits a month and premium models, and Copilot Max ($100) gives 20,000 credits with priority access to premium models. ChatGPT Pro costs $100, $200 or $500 a month, with expanded Codex use, longer sessions and, OpenAI says, currently no five-hour limit. All are billed monthly for new subscribers. For other coding tools, see coding.