AI coding tool prices and usage limits
What Claude Code, Codex, Cursor, Copilot and Gemini CLI cost, and what each vendor's own page says you get for it.
As of October 9, 2026. A script re-reads every page cited here. When one changes, the change is listed at the bottom of this page.
GitHub Copilot and Gemini CLI state a number for what you get. Anthropic, OpenAI and Cursor don't. The five also count in three different units, so their numbers don't compare directly. Each entry below quotes the vendor's own page and names the plan the number belongs to.
AI coding plans compared
Individual plans, US dollars a month. Every entry is from the vendor's own page, as of October 9, 2026.
Claude Code (Anthropic)
Codex (OpenAI)
Cursor
GitHub Copilot
Gemini CLI (Google)What upgrading buys
How many times more you pay, against how many times more the vendor says you get.
Five-hour windows, and no fixed number
Claude Code runs on a five-hour window plus weekly limits, with no message count
Anthropic's pricing page lists Pro at $20 a month, or $17 a month when billed yearly. Max starts at $100 a month and gives "5x or 20x more usage than Pro." Limits reset "on a rolling five-hour session window, and paid plans add weekly limits on top." Chat and Claude Code share them. Usage depends on the model and the conversation, "so there's no fixed message count."
OpenAI's Codex Pro plans come at $100, $200 or $500 a month, with no five-hour limit stated
The Codex pricing page gives ranges of "local messages per five-hour period" for the $20 Plus plan: 15 to 160 with GPT-6.1 Sol, 15 to 150 with GPT-6 Sol. Pro is sold in three sizes, "at $100, $200, or $500 USD per month," and "Pro plans currently have no five-hour limit." The $500 tier adds Ultrafast mode, which uses included usage at 8x the standard rate. For every plan, "Weekly limits may also apply," and the ranges "are not fixed message limits."
Monthly pools and credits
Cursor's $20 Pro plan includes two usage pools and gives a size for neither
Cursor's docs price Pro at $20 a month, Pro Plus at $60 and Ultra at $200. Each includes two pools that reset monthly. One covers Cursor's own models. The other covers third-party models, "charged at the model's API price." The page states no amount for either pool. It says "your model selection affects how quickly your included usage is consumed."
GitHub Copilot's $10 plan includes 1,500 AI credits a month
GitHub's plans page lists Copilot Pro at $10 a month with 1,500 AI credits. That's 1,000 base credits plus a 500 "flex allotment," which the page doesn't define. Pro+ is $39 with 7,000 credits. Copilot Max is $100 with 20,000. Code completions aren't billed in credits. How much work one credit covers is on a separate GitHub billing page.
Requests per day
Gemini CLI allows 1,000 requests a day on a free Google login
Google's quota page lists "Maximum requests per user per day" and no prices. A free Google login gets 1,000 a day. With a free Gemini API key it's 250 a day, on Google's smaller Flash model only. Subscribers to Google AI Pro get 1,500 a day and Google AI Ultra 2,000. Per-minute limits also apply, with no number given.
Changes we've recorded
October 9, 2026: Claude's pricing page now lists the Max and Team API credits
Claude's pricing page now shows the monthly API credits from the October 7 entry below. The Max card adds "$100 to $200/month in API credits to build on the Claude Platform," and the Team card adds "Up to $500 per month in pooled API credits." A new "New Monthly API credits" row in the plan tables reads No for Free and Pro, $100/month for Max 5x and $200/month for Max 20x. For business plans it reads "Up to $500/month" for Team and No for both Enterprise options. Each of these carries a † mark. The page's menu also gained a "Financial advisors" link. No price or limit changed.
October 8, 2026: Codex adds Ultrafast for GPT-6.1 Sol, billed at 8x included usage
The Codex pricing page adds a row for GPT-6.1 Sol Ultrafast. Like GPT-6 Astra Ultrafast, it uses included subscription usage at 8x the Standard rate and purchased credits and Enterprise pay-as-you-go usage at 6x. "Astra Ultrafast" on the $500 Pro tier is now "Ultrafast mode." The same edit deleted the sentence "These billing multipliers don't describe speed increases." OpenAI's speed page puts Ultrafast on "Pro $500 and eligible Enterprise and Edu plans" and says "Other self-serve plans don't have access to Ultrafast at launch, even with purchased credits." No plan price changed.
October 8, 2026: Claude Docs, Slides and Design reach the Free plan, and Dashboards arrives on paid plans
Claude's pricing page now lists Claude Design, Claude Slides and Claude Docs as separate rows, each included on Free, Pro and both Max tiers. The Free column said "No" before. It adds Claude Dashboards (beta) on Pro and Max, and Claude Motion (beta), which the individual-plan table marks unavailable on all four. Anthropic's announcement puts Motion on Team and Enterprise. The Pro FAQ no longer names Design, Slides and Docs among Pro's features. No price or limit changed.
October 7, 2026: Max and Team plans get monthly API credits
Anthropic's help article says Max and Team plans now include monthly credits for the Claude API: $100 on Max 5x, $200 on Max 20x, and on Team $20 per Standard seat and $100 per Premium seat, pooled and capped at $500. Free, Pro and Enterprise plans don't get them. The credits don't cover interactive Claude Code in the terminal, IDE, desktop or web, or extra usage, and they "don't change your usage limits in Claude, Claude Code, or Claude Cowork." Unused credits expire at the end of each billing cycle. The plans' prices did not change, and Claude's pricing page did not mention the credits when we checked on October 8 at 04:33 UTC.
October 7, 2026: Anthropic adds Claude Haiku 5.5 and halves Sonnet 5.5's cache-read price
Claude's pricing page now lists Haiku 5.5 where it listed Haiku 4.5: $0.10 per million input tokens and $0.50 per million output for prompts up to 100K tokens, with cache reads at $0.01 and cache writes at $0.125. Prompts over 100K tokens cost five times as much ($0.50 input, $2.50 output). Haiku 4.5 was $1 input and $5 output. The same page lowered Sonnet 5.5's cache-read price from $0.20 to $0.10 per million tokens. Its input, output and cache-write prices did not change. Cursor's model table carries both changes. GitHub's plans page added Claude Haiku 5.5 to its model list. The Claude Code costs page now names Haiku 5.5 among the models where thinking can't be turned off. It also says the page covers Claude Code usage only and points to the Help Center for other products' limits.
October 5, 2026: Copilot's plans page dropped four retired models
GitHub's plans page no longer lists Claude Opus 4.7, Gemini 3.5 Flash, Gemini 3.6 Flash or Kimi K2.7 Code, which GitHub retired on October 2. No price or limit changed.
October 2, 2026: Codex's per-task credit estimate for GPT-5.6 Sol fell to "2-15"
The Codex pricing page said "A typical GPT-5.6 Sol task may use 5-30 credits." It now says "2-15 credits." The same edit says GPT-6.1 Sol "is available in Work and Codex, not Chat," and that for Enterprise and Edu it "is off by default until an administrator enables it." The per-credit rates did not change.
October 2 to 6, 2026: wording changes on Anthropic's pages
The Claude Code costs page now points to "Organization settings > Usage" where it said "Admin settings > Usage," and dropped a note about /compact in a fresh session. The claude.com pricing page gained "Frontier Academy" and "Campus" links in its menu. On October 6 the costs page also dropped a note that versions before v2.1.239 did not apply the 1.1× data-residency rate to the session cost figure. No price or limit changed.
September 30, 2026: OpenAI's Codex Pro went from "5x or 20x" to three priced tiers
The pricing page said "5x or 20x more Codex usage than Plus" and gave message ranges for "Pro 5x" (70 to 700 with GPT-6 Sol) and "Pro 20x" (300 to 3,000). It now says "Plans at $100, $200, or $500 USD per month" and "Pro plans currently have no five-hour limit." The Pro columns are gone. A June referral promotion was removed too.
September 30, 2026: GPT-6.1 Sol appeared on the Codex and Copilot pages
A new model row on the Codex page, 15 to 160 messages per five hours on Plus, and a new entry in Copilot's model lists. No price or limit changed on Copilot's page.
September 30, 2026: Anthropic added a paragraph to the Claude Code costs page
It describes the /insights report. No price or limit changed.
The New Way is written with AI. It gathers the day's stories, checks them against their sources and drafts every summary. A person decides what runs and reviews every issue before we hit send.