Claude Pro vs ChatGPT Plus for coding in 2026: which $20 plan to buy

Disclosure: Enjoy is our product, and it runs Claude Code, Codex and Antigravity on each company’s plans. We checked every plan and price below against Anthropic’s, OpenAI’s, Google’s and GitHub’s own pages, and the figures against Artificial Analysis’s and SemiAnalysis’s, on November 8, 2026, and we say where you don’t need Enjoy at all, which is most of the time.

Claude Pro vs ChatGPT Plus, compared

At $20 a month, each lab’s plan pays for its own coding agent. “Claude Code is included in all paid plans,” says Claude’s pricing, and OpenAI’s Codex pricing lists “Codex on the web, in the CLI, in the IDE extension, and on iOS” under Plus. So choosing between them mostly means choosing an agent, and how its limits are counted. Google and GitHub sell the nearest alternatives, and some people pay for two.

Plan	Best for	Price, checked November 8, 2026	Agent and top models	Limits	Runs on
Claude Pro	The hardest work, in Claude Code	$20 a month, or $17 billed yearly	Claude Code: Opus 5.5 by default; Fable 5.1 on usage credits	Five-hour and weekly, shared with your Claude chats	Mac, Windows, Linux
ChatGPT Plus	A lower cost per task, and GPT-6 Astra	$20 a month	Codex: GPT-6.1 Sol, GPT-6 Astra, GPT-6.1 Terra, GPT-6 Luna	Five-hour estimates per model, and weekly; shared with ChatGPT from December	Mac, Windows, Linux; web and iOS
Google AI Pro	Gemini and Claude’s models in one Google plan	$19.99 a month, or $199.99 a year	Antigravity: Gemini 4 Argon; Claude Sonnet 5.5 and Opus 5.5 on paid, non-trial plans	Refreshed every five hours, until a weekly limit	Mac, Windows, Linux
GitHub Copilot Pro	Writing code yourself, at half the price	$10 a month; Pro+ $39	Copilot: Claude Sonnet 5.5, GPT-6.1 Terra; Opus 5.5 and GPT-6.1 Sol from Pro+	1,500 AI credits a month, worth $15	Your editor, GitHub.com and a CLI
Claude Pro and ChatGPT Plus	Two labs’ limits, and a second opinion	$40 a month, or $37 with Claude billed yearly	Both agents	Each plan’s own	Mac, Windows, Linux
Enjoy, with the first three	One conversation across your plans	A monthly subscription on top of your plans, free for 7 days (pricing)	Claude Code, Codex, Antigravity with Gemini models, and other labs’ agents	Your plans’	Mac, Windows, Linux

Prices are in US dollars, before tax, as each company lists them. Copilot’s credits are worth a cent each: “1 AI credit = $0.01 USD,” say GitHub’s billing docs.

Which plan gives more coding

Neither lab’s pages can settle it. Anthropic says “there’s no fixed message count,” and OpenAI that “Your allowance does not correspond to a fixed number of messages or tasks.” OpenAI does publish estimates for Plus: 15 to 160 local Codex messages every five hours with GPT-6.1 Sol, 5 to 45 with GPT-6 Astra and 350 to 3,000 with GPT-6 Luna, adding “These estimates are not fixed message limits” and “Weekly limits may also apply.” Anthropic gives Pro only as a multiple: “at least 5x more usage per 5-hour session than Free.”

Two outside measures pull in opposite directions:

• Claude’s plans have measured more API value. SemiAnalysis watched each plan’s meters and found that at the everyday models, Opus 5.5 against GPT-6.1 Sol, Anthropic was “offering ~5x the API-equivalent value across the board,” and that “Plus is also comparable for Sol but offers a relatively worse deal on Astra.” That measure dates from October 5, and Claude subscribers have reported smaller weeks since mid-October.
• Codex spends fewer of those dollars per task. On Artificial Analysis’s comparison, a benchmark task costs $11.42 in API prices on Claude Code with Opus 5.5 at max effort, using 14.5 million tokens, and $1.50 on Codex with GPT-6.1 Sol at max effort, using 3.94 million. SemiAnalysis warns that “Token efficiency is also an extremely relevant factor, but the industry unfortunately lacks reliable data here.”

So how much of your own work fits in each is something only a week of it can show. Both labs show what’s left in Settings > Usage.

How the two differ

	Claude Pro	ChatGPT Plus
Coding Agent Index, best setup	62, Claude Code with Opus 5.5 at max effort	55, Codex with GPT-6 Astra at max effort; 54 with GPT-6.1 Sol
API cost and time per task on the index	$11.42 and 1.0 hour	$7.04 and 24.8 minutes with Astra; $1.50 and 23.7 minutes with Sol
Top model on the plan	Opus 5.5; Fable 5.1 only on usage credits	GPT-6 Astra, at 5 to 45 local messages every five hours
What shares the limit	All of Claude, your chats included	Codex and ChatGPT Work; from December, ChatGPT too, apart from Instant chats
Published limits	A five-hour and a weekly limit, without counts	Five-hour estimates per model; weekly limits “may also apply”
After the limit	Usage credits, at standard API rates	Credits
Billing	Monthly, or $200 a year up front	Monthly

The index’s scores come from three benchmarks, FrontierCode v1.1, Terminal-Bench 4.0 and SWE-Atlas-QnA-AA, as Artificial Analysis’s page stood on November 8. Its costs are “pay-per-token API cost, not consumer plan pricing”: on a plan, a task draws on your limits instead. Our Claude Code vs Codex guide compares the two agents themselves.

Claude Pro

Best for: the hardest work in Claude Code, in focused sessions.

Claude Code’s model guide says its default on “Pro, Max, Ultra, Team, Enterprise, and Anthropic API” is Opus 5.5, the model behind the index’s top score. Claude Code runs in a terminal on macOS, Windows and Linux, in the Claude desktop app and in your editor, and Pro adds “Fable, paid with usage credits,” with Claude’s chat, Projects and Research.

Price: $20 a month, or $17 a month with “annual subscription discount ($200 billed up front).”

Watch for: one pool on two clocks. “Pro plans have a five-hour limit and a weekly usage limit that applies across all models,” says Anthropic’s Pro page, and its pricing says “your work in the terminal and your chats draw from one pool,” so a long chat leaves less for coding. It adds “we may limit your usage in other ways, such as weekly and monthly caps or model and feature usage.” Past the limit, usage credits bill “at standard API rates,” and Anthropic’s Claude Code article says “Consider upgrading to the Max plan if you consistently hit limits and need more capacity for larger repositories.” Our Pro vs Max guide works out when that pays.

ChatGPT Plus

Best for: a lower cost per task, published estimates of what fits, and GPT-6 Astra at $20.

OpenAI describes Plus as “Power a few focused coding sessions each week.” Its card lists GPT-6.1 Sol and GPT-6 Luna, and OpenAI’s Astra guide says “Plus includes Astra in Work and Codex, but not GPT-6 Pro in Chat.” Codex runs in the ChatGPT desktop app, the CLI and an editor extension, on the web and on iOS, with “Cloud-based integrations like automatic code review and Slack integration.”

Price: $20 a month, “billed monthly,” per OpenAI’s Plus article.

Watch for: the allowance is about to change shape. “Starting in December, Plus, Pro, and Business plans’ usage allowance will be shared across ChatGPT and Codex,” OpenAI’s help center says, measured in tokens, and “Instant chat will not draw from your usage.” It hasn’t named the day or the new size, and says “We expect most users won’t need to wait for their usage to reset more often than they do now,” as our memo explains. Today, “You may reach the five-hour limit before five hours have passed, even if you still have weekly usage remaining,” and Astra’s estimate is 5 to 45 messages a window. Past the limit, Plus users “can purchase additional credits to continue working without needing to upgrade their existing plan.” Our Plus vs Pro guide covers the upgrade.

Google AI Pro, for Antigravity

Best for: Gemini and Claude’s models in one $20 plan, with Google’s storage and apps.

Google’s plans page promises “Expanded rate limits and access to Gemini 4 Argon in Google Antigravity, our agentic development platform,” and Antigravity’s models page adds Claude Sonnet 5.5 and Opus 5.5, “Available on Google AI Pro for non-trial subscriptions only.” Antigravity has an app, an IDE and a CLI, and it downloads for macOS, Windows and Linux. On the index, Antigravity CLI with Gemini 4 Argon scores 53, at $11.73 a task in API prices. The plan also includes 5 TB of storage and the Gemini app.

Price: $19.99 a month, or $199.99 a year.

Watch for: the quota. Pro gets “A high, generous quota, refreshed every five hours until the weekly limit is reached,” says Antigravity’s plans page, and “Usage limits for this service are subject to modification.” Claude’s models need a paid plan, not a trial, and the older Claude Sonnet and Opus 4.6 “Will be removed on November 18, 2026.” Artificial Analysis lists Antigravity CLI with Gemini models only, so Claude in Antigravity has no index score of its own. Our Google AI Pro vs Ultra guide has the details.

GitHub Copilot Pro

Best for: writing the code yourself, with suggestions as you type and an agent beside them, for $10.

Copilot Pro is “$10 USD per month (free for some users),” with 1,500 AI credits a month, 1,000 base and a 500 flex allotment, per GitHub’s plans. Inline suggestions “are unlimited and do not consume credits.” Its models include Claude Sonnet 5.5 and GPT-6.1 Terra. Claude Opus 5.5, GPT-6.1 Sol and GPT-6 Astra start at Pro+, $39 a month with 7,000 credits.

Price: $10 a month; Pro+ is $39.

Watch for: credits last a month, not five hours, and they don’t roll over: “Included AI credits do not carry over between months.” The flex allotment “is a variable part of your included usage; it is designed to adapt as the economics of AI evolve.” At a cent a credit, Pro’s allowance is $15 of model use. Our Copilot vs Claude Code guide compares the two.

Claude Pro and ChatGPT Plus together

Best for: running out now and then, or wanting a second lab’s review of the same code.

Anthropic’s and OpenAI’s limits are separate, so when one agent runs out, the other still works, and $40 is less than half of Claude Max, $100 for five times Pro’s usage. One file can instruct both: Claude Code “can read AGENTS.md as your project instructions,” its docs say, “so a repository already set up for other coding agents works without adding a CLAUDE.md, an import, or a setting.”

Price: $40 a month, or $37 with Claude Pro billed yearly.

Watch for: two five-hour limits, and each agent keeps its own conversations, so you carry the context across yourself unless a tool does it for you. Our guide to using both compares ten ways.

Enjoy, with the first three

Best for: paying for two of these plans and moving one conversation between their agents.

Enjoy isn’t an AI plan. It’s a desktop app for Mac, Windows and Linux that runs Claude Code, Codex and Antigravity, with other labs’ agents and local models through OpenCode, on the plans you already pay for. What it adds to this choice:

• Both meters side by side. Its usage meters show each signed-in account’s limits and when each resets, so you can run the week of your own work this guide recommends on both plans at once.
• One conversation, either agent. Switch agents at any point: “The new agent gets the conversation so far, including replies from other agents, before it works on your message.” When one plan runs out, a waiting conversation can continue with another agent.
• Accounts handled. With more than one Claude Code or Codex account signed in, Enjoy switches to one with usage left when another runs out, or waits for the reset and carries on.

Price: Enjoy is one monthly subscription, and the first seven days are free, with a card to start. It includes your phone and any browser and one computer; each other computer is $9.99 a month, and each teammate you invite needs a subscription of their own. See pricing.

Watch for: Enjoy doesn’t change any plan’s limits or add one lab’s models to another’s plan, and it doesn’t run Copilot. It runs Antigravity with Gemini models, so Google AI Pro’s Claude models stay in Google’s own apps. Enjoy runs Claude Code headless, so if the reports in issue #97074, still open, are right, Claude Code in Enjoy uses Pro’s five-hour window faster than in a terminal. Waiting for a reset needs “Your computer must be on, with Enjoy open, when the usage resets,” and scheduled workflows run only while it’s awake with Enjoy open. Enjoy isn’t open source, and it’s a monthly subscription on top of your AI plans once the free week ends.

When to stay on the plan you have

If you already pay for one of these and its agent does your work, stay. An agent you trust is worth more than another lab’s numbers, and every plan here can be bought by the month, so you can try another for a month when you have a reason. Don’t move over December’s change yet: OpenAI hasn’t published the allowance that would show what it means.

If you write most of your code yourself, Copilot Pro’s unlimited suggestions may matter more than either agent’s limits. And if you run out most weeks, compare a second $20 plan with the plan above yours: our guides to Claude’s plans, ChatGPT’s and the $100 and $200 tiers work out when an upgrade pays.

How we compared

We read Anthropic’s pricing, its help center articles on Pro and on Claude Code with Pro, and Claude Code’s model, setup and instructions docs; OpenAI’s Codex pricing and its help center articles on Plus, Astra and the December change; Google’s AI plans and Antigravity’s plans, models and download pages; GitHub’s Copilot plans and billing docs; Artificial Analysis’s index and two of its comparisons; and the public part of SemiAnalysis’s October 5 article, all on November 8, 2026, and every one is linked below. We didn’t measure any plan ourselves. We quote each company’s own words for its limits and leave out what it hasn’t published, including message counts for Claude and the size of December’s allowance. Plans and limits change often, so we recheck this guide against its sources at least once a month, and the date at the foot of the page says when we last did.