Codex vs GitHub Copilot

OpenAI Codex is a coding agent included with ChatGPT plans, while GitHub Copilot is GitHub's assistant with completions, a coding agent and models from several makers, from $10 a month; as of Nov 6, 2026 Copilot is cheaper and lives where code is hosted, and its Pro+ plan can even run Codex as an agent, while Codex gives more agent use per dollar with OpenAI's own models.

GitHub Copilot is the cheaper subscription, $10 a month for Pro against $20 for ChatGPT Plus with Codex, and the two overlap more than ever: both assign a task to an agent that returns a pull request, both review pull requests, and Copilot's Pro+ plan at $39 a month lists Codex itself as one of its third-party agents. The difference is in what the money buys. Copilot meters agent work in GitHub AI Credits and offers models from Anthropic, OpenAI, Google, xAI and Microsoft, plus Kimi K3; Codex comes with ChatGPT, runs OpenAI's GPT-6 models and states its allowance in messages per five hours. The two also share a name in history: the 2021 OpenAI Codex model powered the first GitHub Copilot. Learn more: https://ai.vi.ki/github-copilot.

Quick facts

Free plan	Codex: ChatGPT Free with GPT-6 Luna in the desktop app; Copilot: Free with 2,000 completions a month and no coding agent, as of Nov 6, 2026
Cheapest plan with a cloud agent	Codex: ChatGPT Plus, $20 a month; Copilot: Pro, $10 a month
Team price	Codex: ChatGPT Business $20 per user billed annually; Copilot: Business $19, Enterprise $39 per seat
Models	Codex: OpenAI's GPT-6 family; Copilot: Claude, GPT, Gemini, Grok, MAI and Kimi K3, on all four personal plans
Usage unit	Codex: messages per five hours, then ChatGPT credits; Copilot: GitHub AI Credits, 1 credit = $0.01
Students	Copilot: Copilot Student, $0 for verified students; Codex: no student plan found

Verdict by task

Task	Codex	GitHub Copilot	Verdict
Price to get a cloud agent	$20 (Plus)	$10 (Pro)	Copilot
Inline completions	None	Unlimited on Pro	Copilot
Choice of models	OpenAI only	Six makers	Copilot
Agent use at $20 or less	Up to 160 GPT-6.1 Sol messages per five hours	$15 of credits a month on Pro	Codex appears larger, but units differ
Working outside GitHub	Desktop app, CLI, Slack, Linear, GitLab	Built into GitHub	Codex
Model quality	Same OpenAI models, or Claude in Copilot	Same	Tie: Copilot offers the leaders of both makers
Students	No plan found	Free	Copilot

Coding agent

Both now hand back a pull request. Copilot's coding agent lets you "assign work, get a PR" on Pro, Pro+ and Max (and the cloud agent on Student too), but not on Free. Codex runs the same kind of task in Codex Cloud, where each task has its own workspace and keeps working while your computer is asleep, on Plus and higher; Codex Cloud is not on Free or Go. Codex also works on your own machine through the ChatGPT Desktop App, the Codex CLI and the Codex IDE Extension.

Copilot Pro+ adds "third-party agents Claude and Codex", so a Pro+ subscriber can send a task to Codex without a ChatGPT plan. GitHub's plans page contradicts itself on whether Pro has them too: its feature list says yes and its table says no.

Which has better models?

Neither, on model quality alone, because Copilot can use OpenAI's models as well. Its model list is the same on all four personal plans and spans Anthropic Claude (including Opus), OpenAI GPT, Google Gemini, xAI Grok, Microsoft MAI and Kimi K3. That matters because the Arena WebDev leaderboard of Oct 27, 2026 puts Claude Opus 5.5 first at 1810, ahead of GPT-6 Astra at 1773, and Copilot can run both. Codex runs only OpenAI's models with a ChatGPT sign-in (Codex Models).

Price and usage

As of Nov 6, 2026, GitHub prices Copilot Pro at $10 a month with $15 of AI Credits, Pro+ at $39 with $70, and Max at $100 with $200; GitHub's docs give the same as 1,500, 7,000 and 20,000 credits. Business is $19 a seat and Enterprise $39. Completions are unlimited from Pro up; Free gets 2,000 a month.

Codex comes with ChatGPT: Free $0, Go $8, Plus $20, Pro $100, $200 or $500, Business $20 per user billed annually. On Plus, OpenAI estimates 5 to 45 GPT-6 Astra messages, 15 to 160 GPT-6.1 Sol messages or 350 to 3,000 GPT-6 Luna messages every five hours, and Pro has no five-hour limit. The ChatGPT plan also brings chat, images, voice and ChatGPT Work.

Verdict: Copilot for the lowest bill and for completions; Codex for heavy agent use on one plan. The units differ (dollars of credits against estimated messages), so no exact comparison exists. Codex details are on Codex Pricing and Limits.

Code review

2:20 Copilot Pro and up include code review; Free does not. Codex reviews when someone comments @codex review on a pull request or when automatic review is on, flagging only P0 and P1 issues, and reads review rules from AGENTS.md (Codex Code Review). GitLab merge requests are in preview in Codex. No published test compares the two reviewers: no reliable comparison yet.

Privacy

On personal ChatGPT plans data sharing for training is on by default and can be turned off; Codex on Business has no training on business data by default. GitHub's data policy for Copilot was not read for this page, so the two are not compared on privacy (Privacy and Data Controls).

Which should you choose?

Choose GitHub Copilot for completions as you type, the lowest price, a free student plan, or the choice of Claude and other models in one place, including Codex on Pro+. Choose Codex if you already pay for ChatGPT, want more agent work at $20, or work outside GitHub in Slack, Linear, GitLab or a terminal. See also Codex vs Claude Code and Codex vs Cursor. Learn more: https://ai.vi.ki/github-copilot.

Frequently asked questions

Is GitHub Copilot the same as Codex? No, but they are related: OpenAI's original 2021 Codex model powered the first GitHub Copilot. Today Codex is OpenAI's coding agent in ChatGPT, and Copilot is GitHub's product with models from several makers.

Can I use Codex in GitHub Copilot? Yes on Copilot Pro+ ($39 a month), which lists Codex and Claude as third-party agents. GitHub's plans page is inconsistent on whether Pro includes them.

Which is cheaper, Codex or Copilot? Copilot Pro costs $10 a month and ChatGPT Plus, the cheapest plan with Codex Cloud, costs $20. Codex runs locally on ChatGPT Free and Go, while Copilot Free has no coding agent.

Does Copilot use GPT models? Yes. Copilot's model list includes OpenAI GPT models alongside Claude, Gemini, Grok, Microsoft MAI and Kimi K3 on every personal plan.

Is the Copilot coding agent better than Codex? No published test compares them as of Nov 6, 2026. With the same model the results depend mostly on that model; Copilot can also run Claude Opus 5.5, which leads Arena WebDev.