Every plan below has a ceiling. Some vendors print it as a number of credits. Most print a multiple of another plan, a window ("every 5 hours", "refreshes weekly"), or no number at all. This page puts the ceilings side by side: what each plan costs, what it includes in the vendor's own unit, and what happens when you run out. Each row links to the vendor page it came from.
Price alerts
Email me when any of these tools change their limits or prices
These limits change without much notice. Get one short email when one of them moves.
Plan prices are USD per month before tax, billed monthly unless noted. "When you run out" is what the vendor says happens after the included usage is gone.
Coding agents and AI IDEs
Coding agents and AI IDEs: plan limits and what happens when you run out
Usage multipleFormerly Max 20x. 10x Pro usage every week; no 5-hour limit for verified accounts (unverified accounts get a temporary limit until verified); separate weekly Fable limit; $200/mo API credits; billed monthly only
Wait for the reset, upgrade, or turn on usage credits at standard API rates
Usage multipleFormerly Max 20x. 10x Pro usage every week; no 5-hour limit for verified accounts (unverified accounts get a temporary limit until verified); separate weekly Fable limit; $200/mo API credits; billed monthly only
When you run out
Wait for the reset, upgrade, or turn on usage credits at standard API rates
New subscriptions are bundled with Grok, so one weekly pool covers the editor and Grok.
Gotchas
The weekly pool is shared with Grok. Plus only includes Grok models.
Plus and Free do not include third-party models. Claude, GPT and Gemini on higher plans are billed at the model's API list price, from prepaid credit, starting with the first request. Cursor docs
Higher plans are a multiple of the Plus pool, not a published request count.
Teams and Enterprise add a Cursor Token Rate of $0.25 per million tokens on eligible third-party models. Individual plans do not pay it. (Checked 31 Oct 2026 · Cursor docs)
Claude Code is included in every paid Claude plan and shares one usage pool with Claude chat.
Gotchas
Most accounts have a rolling 5-hour limit, and paid plans add a weekly limit on top. You can hit the 5-hour wall with weekly room left.
Anthropic does not print a fixed message count.
Anthropic says it may limit usage in other ways, including weekly and monthly caps or model and feature limits, at its discretion. Claude pricing
A help article updated 31 Oct 2026 says the higher Max plan is now called Ultra, and that unverified accounts keep a temporary limit. The rows in the table are the tracked plans. Ultra help article
After the limit, paid plans can turn on usage credits at standard API rates.
windsurf.com/pricing now redirects to devin.ai/pricing.
Gotchas
Quotas refresh both daily and weekly. No quota size is published for any plan.
Past the included quota, extra usage is billed at API pricing. Devin pricing
GitHub Copilot
Checked 31 Oct 2026 · Not yet tracked daily · Source
Since 1 June 2026, Copilot plans bill in GitHub AI credits. 1 credit is $0.01.
Gotchas
Each paid plan has base credits plus a flex allotment. GitHub says the flex part is variable and designed to adapt as the economics of AI evolve. It can change without the price changing. Copilot billing
The allowance resets to the full monthly amount at 00:00 UTC on the 1st of each month.
Usage past the pool is $0.01 per AI credit. GitHub says additional usage for individual plans may be capped.
Code completions and next edit suggestions stay unlimited on paid plans and do not use credits.
Copilot Pro and Pro+ on an existing annual plan still use premium requests. The legacy article lists 300 per month and 1,500 per month, at $0.04 per request. Legacy request billing
GitHub's 1 June 2026 changelog said new sign-ups for Student, Pro, Pro+ and Max were paused. Check the plans page for current status. GitHub changelog
OpenAI Codex
Checked 31 Oct 2026 · Not yet tracked daily · Source
Codex comes with ChatGPT plans and shares the Work and Codex allowance.
Gotchas
OpenAI publishes a range for Plus, not a single cap: an estimated 15–160 local GPT-6.1 Sol messages per 5 hours. Weekly limits may also apply. Codex pricing
Fast mode uses the included allowance at 2.5x. Ultrafast, listed on Pro 500, uses it at 8x.
Pro plans currently have no 5-hour limit. Each tier still has a weekly allowance, and OpenAI does not print that allowance as a number.
The Pro tiers article, updated 29 Oct 2026, says Pro 100 and Pro 500 doubled their Work and Codex allowance, and a reduction announced for Pro 200 was cancelled. Pro tiers
JetBrains AI
Checked 31 Oct 2026 · Not yet tracked daily · Source
Credits are dollars. 1 AI credit equals $1 USD, charged in your local currency. The rows below are the individual plans.
Gotchas
The quota resets every 30 days from the date you first use the license. JetBrains docs
JetBrains calls this a rough approximation: one credit is about 10 code-generation requests in AI Chat, or about 40 in the editor.
A second table on the same page lists organization prices, and those differ from the individual plans above.
Augment Code
Checked 31 Oct 2026 · Not yet tracked daily · Source
The price is flat for the workspace, not per seat.
Gotchas
The included amount is dollars of usage, not requests. Model inference is billed at the provider's public API list price plus a flat 40% service fee. The fee is not charged on compute. Augment pricing
Usage is pooled across up to 50 seats, so one heavy user can drain the team balance.
Amazon Q Developer
Checked 31 Oct 2026 · Not yet tracked daily · Source
Free prints an agentic-request cap. Pro does not.
Gotchas
Free has a printed limit of 50 agentic requests a month. Pro says included, with limits. Amazon Q pricing
AWS says an account's usage can be adjusted for regional factors, payment history or fraudulent usage.
Cancelling before the end of the month still charges the full month.
Most paid tiers are priced per month and billed annually.
Gotchas
The vendor FAQ says yearly billing saves 20% on every paid plan. Base44 pricing
Why limits change, and why to get an alert
Vendors pay for every token their models generate, and model prices, new models and demand keep moving. Plans move with them, and usually the price stays the same while what is inside it changes.
Claude, late Oct 2026.Anthropic renamed its higher plans. The help article says verified accounts on the renamed plan have no 5-hour limit, and unverified accounts keep a temporary limit. The table still shows the tracked rows.Anthropic help article
ChatGPT Pro, late Oct 2026.The help article, updated 29 Oct 2026, says Pro 100 and Pro 500 doubled their Work and Codex allowance, and a reduction announced for Pro 200 on 29 September was cancelled.OpenAI help article
GitHub Copilot, 1 Jun 2026.Copilot plans started billing in AI credits. Copilot code review started using Actions minutes. The changelog also said new sign-ups for Student, Pro, Pro+ and Max were paused.GitHub changelog
Cursor, Oct 2026.New subscriptions moved onto a weekly pool shared with Grok. See theCursor docs and theBurnRate change log.
Lovable, 31 Oct 2026.The temporary Chat pricing ends on this date, according to Lovable's docs. The docs do not say what replaces it.Lovable docs
Most of these limits are written as multiples, ranges or "limits apply", so a plan can shrink or grow without the price on the card changing. Vendors announce these changes in help-center articles, changelogs and emails to current subscribers, and those are easy to miss if you are not on that plan yet.
BurnRate checks every tracked pricing page daily. A person checks each change before it is published, and each change is logged with its date and source. The email is the same log, sent to you: one short note a week, only when a price, credit grant or limit actually moves. No change, no email. The change log has each one.
Price alerts
Email me when any of these tools change their limits or prices
Know before your plan changes. No change, no email. Free, double opt-in.
Questions
Which AI coding tool gives the most usage for the money?
No one can answer that from the vendor pages, because the units do not match. Cursor sells a multiple of a weekly pool, Claude a multiple of a session window and a weekly window, Copilot and JetBrains dollar-denominated credits, Codex estimated message ranges. Compare what happens when you run out instead: a hard wait, a slower fallback, or metered billing at API rates.
What is a "5-hour limit"?
A rolling window. Claude, and ChatGPT Plus for Codex, count usage within a 5-hour session, and paid plans add a weekly limit on top. You need room in both to keep going. Anthropic's Ultra help article and OpenAI's Pro tiers article describe which higher plans currently have no 5-hour limit. They still have a weekly one. The Claude rows in the table are the tracked plans.
What happens when I hit my limit?
It depends on the vendor. Cursor: wait for the weekly refresh, or pay per model from prepaid credit. Claude: wait, upgrade, or use usage credits at API rates. Copilot: pay $0.01 per AI credit, with a possible cap, upgrade, or wait for the 1st. Codex: buy credits or use an API key. The table's When you run out column has each vendor's own wording.
Why do some rows say "not published"?
Because the vendor does not print the number. BurnRate only records figures stated on the vendor's own pricing or docs pages, and leaves the rest blank rather than estimating.
How often is this page updated?
Tracked tools (Cursor, Claude Code, Windsurf, Lovable, Bolt, Replit, v0, Base44) are checked daily, and the Checked and Verified dates on each row are live. Rows marked "Checked 31 Oct 2026" are tools BurnRate does not track yet. They were last read on that date.
Does BurnRate get paid to list a tool?
Some Try it links are affiliate links and there is one labelled sponsor slot. Neither changes a number or a row's position. See the disclosure page.
Part of BurnRate. Prices can move. Checked is the last daily pass that read them. Verified is the last human check.