AI usage limits: which plan to buy
Claude Pro and ChatGPT Plus ($20) each include an S-tier coding agent with a 5-hour usage window. Cursor and Copilot use monthly or weekly credit pools. App builders like Lovable and Bolt charge by credits or tokens.
Prices, usage limits and what happens when you run out, for 10 AI coding and app-building tools (24 paid plans). Every row links to the official page.
Last updated: · Reviewed weekly against official pricing pages.
Last verified: · plan names, prices and monthly API credits for ChatGPT / Codex re-read on the official pages that day. Monthly prices for Claude (Pro / Max / Ultra) were last compared on 31 Oct 2026. Monthly prices for Cursor and GitHub Copilot were last compared on 28 Oct 2026. Usage-limit wording and the other tools keep the check dates in the table.
Plans and limits compared
| Tool | Plan | Price/mo | What you get | How limits work | When you hit the limit | Source · checked |
|---|---|---|---|---|---|---|
| Claude (Pro / Max / Ultra) SAnthropic | Pro | $20$17/mo billed annually | At least 5× Free usage per 5-hour session, plus a weekly limit. Includes Claude Code. | Rolling 5-hour session limit plus weekly limits on Pro and Max, shared by Claude chat and Claude Code. Verified Ultra accounts have no 5-hour limit, only weekly limits; unverified Ultra accounts keep a temporary limit. No fixed message count: it depends on model, conversation length and features. | Wait for the reset, upgrade, turn on usage credits (billed at standard API rates), or switch Claude Code to pay-as-you-go Console API credits. A response already in progress may get a small 'wrapping up' allowance. | Official Official 2 Official 3 Official 4 |
| Max | $100Monthly billing only. Listed here as Max 5x until this update. | 5× Pro usage every 5 hours and every week. Fable included, up to half your weekly limit. Priority access at high traffic times. Includes $100 a month in API credits (table below). | ||||
| Ultra | $200Monthly billing only. Anthropic's help page says Max 20x is now called Ultra. | No 5-hour limit for verified accounts (unverified accounts keep a temporary limit). 10× Pro usage every week. Fable included, up to half your weekly limit. Includes $200 a month in API credits (table below). | ||||
| ChatGPT / Codex SOpenAI | Go | $8 | Limited Codex: GPT-6 Luna in the desktop app, subject to rollout. | Usage allowance per 5-hour window on Plus; weekly limits may also apply. OpenAI publishes ranges, not fixed limits: on Plus, 5–45 local GPT-6 Astra messages, 15–160 GPT-6.1 Sol, 35–285 GPT-6.1 Terra, 15–150 GPT-6 Sol or 350–3,000 GPT-6 Luna per 5 hours (table below). Cloud tasks may use more, Fast mode uses your included allowance at 2.5×, and image generation uses included limits about 3–5× faster. OpenAI says that starting in December the Plus, Pro and Business allowance will be shared across ChatGPT and Codex and measured in tokens (OpenAI gives no exact date or per-plan token amounts yet). | A turn already running can continue, subject to fair use limits. Plus and Pro can buy ChatGPT credits to keep going, or run local chats with an API key at API rates. | Official Official 2 Official 3 Official 4 Official 5 |
| Plus | $20 | 5-hour window. OpenAI's estimates for local messages per 5 hours: GPT-6 Astra 5–45, GPT-6.1 Sol 15–160, GPT-6.1 Terra 35–285, GPT-6 Sol 15–150, GPT-6 Luna 350–3,000. From December the allowance is shared with ChatGPT chat and measured in tokens. | ||||
| Pro | $100Three tiers: Pro 100 ($100), Pro 200 ($200), Pro 500 ($500). Monthly billing only. | Currently no 5-hour limit on Pro plans. Pro 200 includes more usage than Pro 100, and Pro 500 offers the highest included usage; OpenAI doesn't state multiples. OpenAI's Pro help page says Pro 100 and Pro 500 now include twice their previous Work and Codex allowance, which doesn't guarantee twice as many finished tasks. Among Pro plans, Ultrafast mode is only on Pro 500. Monthly API credits equal the tier price (table below). | ||||
| Cursor SCursor | Plus | $8 | Basic shared pool, Grok models only. | One AI usage pool shared across Cursor and Grok that refreshes weekly. Plus includes Grok models only. Third-party models (Claude, GPT, etc.) are not in the pool: they're billed separately at API list price. | Buy top-offs or wait for the weekly refresh. Spending caps are available for third-party model usage. | Official |
| Premium | $30 | 4× Plus usage. Third-party models at API list price. | ||||
| Super | $100 | 15× Plus usage. Priority model access at peak times. | ||||
| Ultra | $200 | 50× Plus usage. | ||||
| GitHub Copilot AGitHub | Pro | $10 | $15 of AI credits per month ($10 base + $5 variable flex allotment). | Monthly GitHub AI Credits (1 credit = $0.01) for chat, agent mode, code review, cloud agent and CLI. Cost per interaction depends on the model. Code completions don't use credits and are unlimited on paid plans. Free: 2,000 completions and 50 chat requests. | Wait for the monthly reset, set a dollar budget for paid usage, or switch to a cheaper model. Business/Enterprise admins can pause Copilot instead. | Official |
| Pro+ | $39 | $70 of AI credits per month ($39 base + $31 flex). | ||||
| Max | $100 | $200 of AI credits per month in total. | ||||
| Google AI / Antigravity AGoogle | Google AI Pro | $19.99 | More Antigravity rate limits plus a flexible AI credit pool. Exact numbers: Not published. | Antigravity has weekly rate limits on Free; Google AI Pro and Ultra add higher rate limits and a flexible AI credit pool. Exact limits: Not published. | Not published. | Official Official 2 |
| Google AI Ultra | $99.99$199.99 option gives 20× AI Pro limits | 5× AI Pro usage limits (20× on the $199.99 option). | ||||
| Devin Desktop (formerly Windsurf) BCognition | Pro | $20 | Higher quotas, including frontier models, plus Devin Cloud agents. | Usage allowance that refreshes automatically on a daily and weekly basis. Cost per message depends on the model and task. Tab completions are unlimited. Exact quotas: Not published. | Pro can buy extra usage at API pricing. | Official |
| Max | $200 | Significantly higher quotas than Pro. | ||||
| Lovable SLovable | Pro | Not publishedNot published in the fetched page text | Monthly plan credits + 5 daily build credits + 20 Cloud credits a month. Top-ups: $15 per 50 credits. | One credit balance for building, chat, hosting (Cloud) and AI features in your app. Plan mode costs 1 credit per message, Build mode varies (e.g. 0.50 for a small style change). Every plan gets 5 daily build credits. Monthly credits roll over and expire after 2 months. | Building stops, AI features in your live app fail, and the built-in backend pauses shortly after. Pro top-ups cost $15 per 50 credits. You may need to click 'Wake up' to restart Cloud. | Official Official 2 |
| Bolt AStackBlitz | Pro | $25 | Starts at 10M tokens a month, no daily limit. Unused tokens roll over for one month. | Monthly token allowance. Bigger projects use more tokens per message, because the project's files are synced to the AI. Free: 300K tokens a day and 1M a month. | Not published on the pricing page. | Official |
| Teams | $30Per member | Per-member token allotment. Tokens aren't shared between members. | ||||
| Replit AReplit | Core | $20$18/mo billed annually | Up to 30 hours of chat in Free Mode, plus $20 a month toward more powerful models. | Free Mode hours for chat, plus a monthly dollar credit for more powerful models. | Not published on the pricing page. | Official |
| Pro | $100$90/mo billed annually | More Free Mode usage, plus $100 a month toward the most powerful models. | ||||
| v0 AVercel | Plus | $30Per user. The page also shows $90, so check the live page. | $30 of credits a month, plus $2 of free credits each day you log in. | Dollar credits spent at per-token model rates (e.g. v0 Pro: $2 input / $10 output per 1M tokens). Free: 7 messages a day. | Buy additional credits, shared across the team. | Official |
| Business | $100Per user | $30 of credits a month + $2 a day. Training opt-out by default. |
ChatGPT Plus: Codex messages per 5 hours, by model
OpenAI estimates how many local messages fit in one five-hour window, model by model. Checked 1 Nov 2026.
| Model | Plus | Standard Business |
|---|---|---|
| GPT-6 Astra | 5–45 | 5–45 |
| GPT-6.1 Sol | 15–160 | 15–160 |
| GPT-6.1 Terra | 35–285 | 35–285 |
| GPT-6 Sol | 15–150 | 15–150 |
| GPT-6 Luna | 350–3,000 | 350–3,000 |
- OpenAI calls these estimates of local messages in a five-hour window, not guaranteed counts. Cloud tasks may use more allowance, and weekly limits may also apply.
- Pro 100, Pro 200 and Pro 500 currently have no five-hour limit in Work and Codex, but still have an included allowance.
- Model names are copied from OpenAI's tables. OpenAI's pricing page lists all five rows; the Astra help page's table omits GPT-6.1 Terra, and its model list calls the models GPT-5.6 Sol, Terra and Luna. The Astra help page does not explain the differences.
- OpenAI's separate usage page says that starting in December, Plus, Pro and Business usage will be shared across ChatGPT and Codex and measured in tokens. Instant chat will not draw from it. It gives no exact date and no per-plan token amounts.
Monthly API credits included with the top plans
Anthropic and OpenAI now include monthly credits for their developer APIs with their higher-priced plans. They are for building your own apps and agents, not for extra chat or coding-tool usage.
| Plan | Price/mo | API credits | Covers | Does not cover |
|---|---|---|---|---|
| Claude MaxAnthropic | $100 | $100/mo | Claude API, Claude Managed Agents, the Claude Agent SDK and the Console playground. | Claude Code, extra usage in the Claude apps, and Claude Platform on AWS, Amazon Bedrock, Google Cloud or Microsoft Foundry. |
| Claude UltraAnthropic | $200 | $200/mo | Claude API, Claude Managed Agents, the Claude Agent SDK and the Console playground. | Claude Code, extra usage in the Claude apps, and Claude Platform on AWS, Amazon Bedrock, Google Cloud or Microsoft Foundry. |
| ChatGPT / Codex Pro 100OpenAI | $100 | $100/mo | Paid OpenAI API usage billed to the linked organization: Responses API, Agents API, Realtime, Images and built-in tools such as web search. | Codex usage through API keys. |
| ChatGPT / Codex Pro 200OpenAI | $200 | $200/mo | Paid OpenAI API usage billed to the linked organization: Responses API, Agents API, Realtime, Images and built-in tools such as web search. | Codex usage through API keys. |
| ChatGPT / Codex Pro 500OpenAI | $500 | $500/mo | Paid OpenAI API usage billed to the linked organization: Responses API, Agents API, Realtime, Images and built-in tools such as web search. | Codex usage through API keys. |
- Credits expire at the end of each billing cycle and don't roll over. Claude's credits arrive monthly on annual plans.
- There's a 7-day wait: Anthropic says new subscribers can claim after 7 days on an eligible plan, and OpenAI requires a paid Pro subscription that is at least seven days old.
- You claim them by linking an API organization you own (a Claude Console organization or an OpenAI API organization). Included credits are spent before any credits you bought.
- Credits don't raise your chat or coding-tool limits. Anthropic says they don't change your usage limits in Claude or Claude Code, and OpenAI says they don't increase ChatGPT or signed-in Codex limits.
- Team (Claude) and Business (ChatGPT) workspaces get pooled credits per seat, capped at $500 a month. They aren't in the table above.
Which AI has the best usage limits?
There is no single winner, because most vendors don't publish fixed numbers. What the official pages say:
- Claude Pro, Max and Ultra: Pro and Max have a rolling 5-hour session limit plus weekly limits, shared by Claude chat and Claude Code, with no fixed message count. Max is 5× Pro every 5 hours and every week. Ultra has no 5-hour limit for verified accounts and gives 10× Pro every week.
- ChatGPT / Codex: Plus has a usage allowance per 5-hour window (OpenAI's estimates run from 5–45 local messages on GPT-6 Astra to 350–3,000 on GPT-6 Luna), and OpenAI says Pro currently has no five-hour limit (Pro 100, 200 and 500). OpenAI publishes ranges, not fixed limits. From December the Plus, Pro and Business allowance is shared across ChatGPT and Codex and measured in tokens, with no exact date or per-plan amounts published yet.
- Cursor and GitHub Copilot: Cursor has one usage pool that refreshes weekly. Copilot gives monthly AI credits, and code completions are unlimited on paid plans.
- Lovable, Bolt and v0: credits or tokens instead of time windows, so a long bug loop drains your balance rather than locking you out.
Compare the rows in the table above, then see the stacks below for a plan at your budget.
Which plan should I buy?
Sample stacks built from the table above and our coding agent and app builder tier lists.
~$20/mo: Claude Pro ($20)
One S-tier coding agent. Claude Pro includes Claude Code, and the limit resets every 5 hours. ChatGPT Plus ($20, Codex) is the S-tier alternative. Add free tiers (Copilot Free, Antigravity Free, Lovable's 5 daily credits) for overflow.
~$50/mo: Claude Pro ($20) + ChatGPT / Codex Plus ($20) + GitHub Copilot Pro ($10)
Two S-tier agents with separate 5-hour windows, so when one hits its limit you switch to the other. Copilot Pro adds unlimited completions in any editor.
~$200/mo: Claude Max ($100) + ChatGPT / Codex Pro ($100)
For all-day agent work: Claude Max ($100, 5× Pro usage every 5 hours and every week) plus Codex Pro at the $100 tier, which currently has no 5-hour limit. One-plan options at $200: Claude Ultra (10× Pro usage every week, no 5-hour limit for verified accounts), ChatGPT Pro 200, or Cursor Ultra (50× Plus, but third-party models cost extra there).
Building with no-code? Lovable (S-tier) and Bolt Pro ($25) charge by credits or tokens instead of time windows, so a long bug loop drains your balance rather than locking you out for a few hours.
Tips to stretch your limits
- Start a fresh session for each task. Long chats re-send the whole history, so every message costs more.
- Use the smaller model for routine edits and save the frontier model for planning and hard bugs. Copilot and Codex both say cheaper models stretch your allowance.
- Keep instruction files short (AGENTS.md, CLAUDE.md, rules), and switch off MCP servers you aren't using, since each one adds context to every message.
- Plan first, then build. Lovable's Plan mode costs a flat 1 credit per message, and a clear plan cuts down on failed build attempts.
- Stop a bug loop after 2–3 failed fixes. Revert, read the error yourself, and give the agent the exact log instead of asking it to 'try again'.
- Check your meter: /status in Claude Code and Codex CLI, Settings → Usage in Claude, and the usage dashboard in Copilot, Cursor and Lovable.
- Alternate between tools with separate windows: when Claude's 5-hour limit hits, move to Codex, and the other way round.
- Watch for surprise API billing. If ANTHROPIC_API_KEY is set, Claude Code bills the API instead of your plan, and auto top-ups are easy to forget.
Related
- Cheapest AI models: per-token API prices, if you'd rather pay as you go.
- Best AI models: tier lists of the models behind these tools, by job.
- Head-to-head tool comparisons
- AI subscriptions: monthly vs annual prices and what $20 a month gets you.
- Stack picker: a suggested stack for your budget.
- Cheapest AI IDE plans: every plan price, cheapest first.
- Cheapest AI app builder plans: every plan price, cheapest first.
Changelog
- OpenAI's Astra usage page now lists GPT-6 Astra (5–45) and GPT-6 Sol (15–150) local messages per 5 hours on Plus, next to GPT-6.1 Sol (15–160), GPT-6.1 Terra (35–285) and GPT-6 Luna (350–3,000). Added a model table, and OpenAI's note that from December the Plus, Pro and Business allowance is shared across ChatGPT and Codex, measured in tokens.
- Anthropic's pricing page now lists Max at $100 and Ultra at $200, and its Ultra help page says Max 20x is now called Ultra. Plan names, prices, 5-hour and weekly wording updated (Max was listed here as Max 5x). OpenAI's Pro help page lists Pro 100, Pro 200 and Pro 500, with twice the previous Work and Codex allowance on Pro 100 and Pro 500. Added the monthly API credits both companies now include (Claude Max and Ultra, ChatGPT Pro 100, 200 and 500).
- Weekly check against the official pricing pages: no price changes found.
- Page launched: 10 tools, 24 plans, all from official pricing pages.
FAQ
What is the Claude Code 5-hour limit?
Claude Pro and Max have a usage limit on a rolling 5-hour session window, plus weekly limits. Claude chat and Claude Code draw from the same pool. Anthropic doesn't publish a fixed message count. Max gives 5× Pro's usage every 5 hours and every week. Ultra has no 5-hour limit for verified accounts (unverified accounts keep a temporary limit) and gives 10× Pro's usage every week.
What happens when I hit my Claude Code or Codex limit?
Claude: wait for the reset, upgrade, or turn on usage credits at API rates. Codex: a running turn can finish, then Plus and Pro can buy ChatGPT credits or switch to an API key.
Does Codex have a 5-hour limit on ChatGPT Pro?
OpenAI's Codex pricing page says Pro plans currently have no five-hour limit. Plus does, and OpenAI says weekly limits may also apply.
Are Claude, GPT and Gemini models included in Cursor's plans?
No. Cursor's pricing page says third-party models aren't in the plan's usage pool and are billed separately at API list price. Plus includes Grok models only.
Which AI coding plan gives the most for $20 a month?
For agent coding, Claude Pro ($20, Claude Code) or ChatGPT Plus ($20, Codex). Both are S-tier on TierPatch. Their limits aren't published as fixed numbers, so we can't rank them on raw usage.
Do Claude Max and ChatGPT Pro include API credits?
Yes. Claude Max ($100) and Ultra ($200) include $100 and $200 a month in Claude API credits. ChatGPT Pro includes API credits equal to the tier price: $100, $200 or $500 a month. They expire at the end of each billing cycle, you can claim them after a 7-day wait, and they don't raise your chat limits. Claude's credits don't cover Claude Code, and OpenAI's don't cover Codex through API keys.
How many Codex messages do I get on ChatGPT Plus?
OpenAI estimates, per five-hour window of local messages: GPT-6 Astra 5–45, GPT-6.1 Sol 15–160, GPT-6.1 Terra 35–285, GPT-6 Sol 15–150 and GPT-6 Luna 350–3,000. These are estimates, not guarantees; cloud tasks may use more, and weekly limits may also apply.
Will ChatGPT and Codex share one usage limit?
OpenAI says that starting in December, the Plus, Pro and Business usage allowance will be shared across ChatGPT and Codex and measured in tokens, and that instant chat will not draw from it. OpenAI gives no exact date and no per-plan token amounts yet.
How often is this page updated?
Weekly, against each vendor's official pricing page. Last updated: 1 Nov 2026.