AI Usage Limits Comparison

Hover any value for its verification date; full dates and sources are in each plan's section. Spotted a change we haven't recorded? Report a limit change — or grab the raw data as limits.json.

All plans at a glance

Click a plan name to jump to its full tracked limits, notes, and sources.

Plan	Price	Messages / requests	Weekly cap	Context window	Deep research	Agent / tool use
ChatGPT Free OpenAI	Free	~10 messages per 5 hours	—	8K tokens	5 lightweight tasks / month	—
ChatGPT Plus OpenAI	$20/month	160 messages per 3 hours	Up to 3,000 per week	32K tokens	25 tasks / month	—
ChatGPT Pro OpenAI	$200/month	Effectively unmetered	—	128K tokens	125 full + 125 lightweight per month	—
Claude Free Anthropic	Free	Not published	—	200K tokens	—	—
Claude Pro Anthropic	$20/month	~45 messages per 5 hours	Applies across all models	200K tokens	—	Included (share the same usage limits)
Claude Max Anthropic	$100-$200/month	Messages (Max 5x, $100/mo): ~225 messages per 5 hoursMessages (Max 20x, $200/mo): ~900 messages per 5 hours	Weekly usage cap (Max 5x): 5× Pro's weekly allowanceWeekly usage cap (Max 20x): 20× Pro's weekly allowance	200K tokens	—	Included, with priority access at peak times
Gemini Free Google AI	Free	Standard compute allowance	—	32K tokens	Not published	—
Google AI Pro / Gemini Advanced Google AI Pro	$19.99/month	4× the standard allowance	—	1M tokens	Not published	Not published
Google AI Ultra Google AI Ultra	$249.99/month	5×–20× the AI Pro allowance	—	1M tokens	Not published	—
Perplexity Free Perplexity	Free	~5 per day	—	—	—	—
Perplexity Pro Perplexity	$20/month	300+ per day	—	—	Not published	—
Cursor Hobby / Free Cursor	Free	—	—	—	—	Not published
Cursor Pro Cursor	$20/month	—	—	—	—	Set monthly model-usage allowance
GitHub Copilot Free GitHub Copilot	Free	Not published	—	—	—	2,000 per month
GitHub Copilot Pro GitHub Copilot	$10/month	1,500 AI Credits per month	—	—	—	Unlimited

✓ Figures verified between Jan 15, 2026 and Jul 29, 2026 against official provider pages · hover any value for its date · full per-fact dates and sources in the plan sections

Recent limit changes

From the full changelog of tracked plan changes.

Jul 29, 2026All plans
Full verification pass against live provider pages: ChatGPT Plus 160/3h and 3,000 Thinking/week reconfirmed; Claude Pro's ~45 messages/5h and all-model weekly cap reconfirmed; Gemini and Copilot facts updated for their 2026 metering changes below.
Jun 1, 2026Copilot Pro
GitHub replaced Copilot's monthly premium requests with GitHub AI Credits: Pro now includes 1,500 credits (≈$15 of model usage) per month, metered by token consumption at each model's rate.
Before: 300 premium requests / month ($0.04 per extra). After: 1,500 AI Credits / month (1 credit = $0.01).
Source
May 15, 2026Gemini Free
In May 2026 Google replaced Gemini's fixed daily prompt counts (5/day free, 100/day AI Pro, 500/day Ultra) with compute-based limits that refresh every 5 hours inside a weekly cap; AI Pro is now '4× standard' and Ultra '5×–20× Pro'.
Before: Fixed daily prompt counts per tier. After: Compute-based allowance, 5-hour refresh + weekly cap.
Source

ChatGPT Free limits

Free · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Flagship-model messages	~10 messages per 5 hours Applies to the flagship model; once the cap is hit, chats continue on the lighter mini model. Varies with demand.	✓ Jan 15, 2026	Official page
Context window	8K tokens ChatGPT-interface context for free accounts. API limits are different and larger.	✓ Jan 15, 2026	Official page
Deep research tasks	5 lightweight tasks / month Free tier runs the lightweight version of deep research only.	✓ Jan 15, 2026	Official page
Image generation	Not published Free image generation is capped daily, but OpenAI has not published a stable number.	✓ Jan 15, 2026	Official page
At the limit	Falls back to a lighter model until the window resets	✓ Jan 15, 2026	Official page

ChatGPT Plus limits

$20/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Flagship-model messages	160 messages per 3 hours Rolling 3-hour window; overflow falls back to the mini model. OpenAI blocks automated access to its help center, so this was cross-checked against multiple current sources citing OpenAI's docs (Jul 2026).	✓ Jul 29, 2026	Official page
Reasoning (Thinking) messages	Up to 3,000 per week Weekly cap on the extended-reasoning model; raised from 200/week in August 2025. Cross-checked against multiple current sources citing OpenAI's docs (Jul 2026).	✓ Jul 29, 2026	Official page
Context window	32K tokens ChatGPT-interface context for Plus accounts.	✓ Jan 15, 2026	Official page
Model access	Flagship + reasoning models, plus legacy model picker	✓ Jan 15, 2026	Official page
Deep research tasks	25 tasks / month Mix of full and lightweight runs; the exact split has changed over time.	✓ Jan 15, 2026	Official page
Image generation	Not published Higher priority and volume than Free, but no fixed published cap.	✓ Jan 15, 2026	Official page
File uploads	80 uploads per 3 hours Cross-checked against multiple current sources citing OpenAI's docs (Jul 2026); OpenAI blocks automated access to its help center.	✓ Jul 29, 2026	Official page
At the limit	Falls back to a lighter model or waits for the window reset	✓ Jan 15, 2026	Official page

ChatGPT Pro limits

$200/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Flagship-model messages	Effectively unmetered OpenAI describes Pro as unlimited subject to abuse guardrails.	✓ Jan 15, 2026	Official page
Context window	128K tokens Largest ChatGPT-interface context tier.	✓ Jan 15, 2026	Official page
Model access	Everything in Plus + Pro-grade reasoning model	✓ Jan 15, 2026	Official page
Deep research tasks	125 full + 125 lightweight per month	✓ Jan 15, 2026	Official page
At the limit	Soft guardrails against automated or abusive use only	✓ Jan 15, 2026	Official page

Claude Free limits

Free · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Messages	Not published Anthropic publishes no fixed free number; capacity varies with demand, message length, and attachments within 5-hour windows. Pro is described as at least 5× free usage.	✓ Jul 29, 2026	Official page
Context window	200K tokens	✓ Jul 29, 2026	Official page
Model access	Latest default Claude model Model picker options change; premium models are reserved for paid tiers.	✓ Jan 15, 2026	Official page
At the limit	Hard stop until the 5-hour window resets	✓ Jul 29, 2026	Official page

Claude Pro limits

$20/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Messages	~45 messages per 5 hours Anthropic's guidance: expect at least 45 messages every 5 hours for shorter conversations — varies with message length, attachments, model, and demand.	✓ Jul 29, 2026	Official page
Weekly usage cap	Applies across all models Added August 2025 on top of 5-hour windows; exact numbers are not published. Resets at a fixed time each week (shown in Settings → Usage). Mainly felt by heavy Claude Code use.	✓ Jul 29, 2026	Official page
Context window	200K tokens	✓ Jul 29, 2026	Official page
Claude Code & Cowork	Included (share the same usage limits)	✓ Jul 29, 2026	Official page
At the limit	Hard stop until the 5-hour window resets; separate weekly cap resets at a fixed weekly time	✓ Jul 29, 2026	Official page

Claude Max limits

$100-$200/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Messages (Max 5x, $100/mo)	~225 messages per 5 hours Derived from Anthropic's published 5× multiplier on Pro's ~45-message guidance; same variability rules as Pro.	✓ Jul 29, 2026	Official page
Messages (Max 20x, $200/mo)	~900 messages per 5 hours Derived from Anthropic's published 20× multiplier on Pro's ~45-message guidance; same variability rules as Pro.	✓ Jul 29, 2026	Official page
Weekly usage cap (Max 5x)	5× Pro's weekly allowance Weekly caps apply across all models; exact numbers are not published. The hour-range estimates Anthropic posted in Aug 2025 (~140–280 h Sonnet-class) are no longer listed.	✓ Jul 29, 2026	Official page
Weekly usage cap (Max 20x)	20× Pro's weekly allowance Weekly caps apply across all models; exact numbers are not published. The hour-range estimates Anthropic posted in Aug 2025 (~240–480 h Sonnet-class) are no longer listed.	✓ Jul 29, 2026	Official page
Context window	200K tokens	✓ Jul 29, 2026	Official page
Claude Code & Cowork	Included, with priority access at peak times	✓ Jul 29, 2026	Official page

Gemini Free limits

Free · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Usage allowance	Standard compute allowance Since May 2026 Google uses compute-based limits: the allowance refreshes every 5 hours until a weekly cap, and prompt complexity, features, and chat length all count. Numeric prompt counts are no longer published (previously 5 top-model prompts/day).	✓ Jul 29, 2026	Official page
Context window	32K tokens	✓ Jul 29, 2026	Official page
Deep Research reports	Not published Available on the free tier but may be unavailable during high demand; numeric caps are no longer published (previously 5/month).	✓ Jul 29, 2026	Official page
Image generation	Not published Image and video generation draw from the shared compute allowance and 'consume more of your usage'; per-day counts are no longer published (previously up to 100/day).	✓ Jul 29, 2026	Official page
At the limit	Allowance refreshes every 5 hours, inside an overall weekly cap	✓ Jul 29, 2026	Official page

Google AI Pro / Gemini Advanced limits

$19.99/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Usage allowance	4× the standard allowance Compute-based since May 2026: refreshes every 5 hours until a weekly cap. Numeric prompt counts are no longer published (previously 100 top-model prompts/day).	✓ Jul 29, 2026	Official page
Context window	1M tokens Google cites roughly 1,500 text pages.	✓ Jul 29, 2026	Official page
Deep Research reports	Not published Included with higher availability than free; numeric caps are no longer published (previously 20/day).	✓ Jul 29, 2026	Official page
Image generation	Not published Draws from the shared compute allowance; per-day counts are no longer published (previously up to 1,000/day).	✓ Jul 29, 2026	Official page
Video generation	Not published Video generation consumes the shared compute allowance (previously ~3 fast-tier generations/day). Bundle also includes 2TB storage and NotebookLM higher limits.	✓ Jul 29, 2026	Official page

Google AI Ultra limits

$249.99/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Usage allowance	5×–20× the AI Pro allowance Depends on the Ultra subscription tier. Compute-based since May 2026: refreshes every 5 hours until a weekly cap (previously 500 top-model prompts/day).	✓ Jul 29, 2026	Official page
Context window	1M tokens	✓ Jul 29, 2026	Official page
Model access	Highest tier, including Deep Think reasoning and agent features; 30TB storage bundle	✓ Jul 29, 2026	Official page
Deep Research reports	Not published Highest availability tier; numeric caps are no longer published (previously 200/day). Deep Think reasoning mode is Ultra-only.	✓ Jul 29, 2026	Official page

Perplexity Free limits

Free · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Pro searches	~5 per day Quick searches are unlimited. Perplexity adjusts the free Pro-search allowance over time; ~5/day is the consistently reported current figure (cross-checked Jul 2026).	✓ Jul 29, 2026	Official page
File uploads	3 per day	✓ Jan 15, 2026	Official page
At the limit	Falls back to unlimited quick searches until the daily Pro allowance resets	✓ Jul 29, 2026	Official page

Perplexity Pro limits

$20/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Pro searches	300+ per day Quick searches remain unlimited (cross-checked Jul 2026).	✓ Jul 29, 2026	Official page
Model access	Choice of frontier models (GPT, Claude, Gemini, Sonar)	✓ Jan 15, 2026	Official page
Research / Labs quotas	Not published Perplexity has not published stable quotas for its research and Labs features.	✓ Jan 15, 2026	Official page
File uploads	Unlimited (fair use)	✓ Jan 15, 2026	Official page

Cursor Hobby / Free limits

Free · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Model access	Limited agent requests and tab completions	✓ Jul 29, 2026	Official page
Agent requests	Not published Hobby lists 'limited Agent requests' and Composer access without numbers (an earlier published figure was 2,000 completions/month).	✓ Jul 29, 2026	Official page

Cursor Pro limits

$20/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Model access	All supported frontier coding models	✓ Jul 29, 2026	Official page
Included model usage	Set monthly model-usage allowance Cursor: 'Every plan includes a set amount of model usage'; community estimates put Pro's allowance around $20 of API-rate usage. Pro+ (3× Pro) and Ultra (20× Pro) tiers extend it. Tab completions are unlimited.	✓ Jul 29, 2026	Official page
At the limit	Continue via on-demand usage, billed in arrears	✓ Jul 29, 2026	Official page

GitHub Copilot Free limits

Free · Official pricing page · Full plan guide

Limit	Value	Verified	Source
Chat / agent usage	Not published GitHub now lists 'limited chat and agent usage' without a number (previously 50 premium requests/month before the June 2026 AI Credits change).	✓ Jul 29, 2026	Official page
Model access	Base model; limited premium-model access	✓ Jan 15, 2026	Official page
Code completions	2,000 per month	✓ Jul 29, 2026	Official page

GitHub Copilot Pro limits

$10/month · Official pricing page · Full plan guide

Limit	Value	Verified	Source
AI Credits (premium usage)	1,500 AI Credits per month ≈$15 of model usage (1 credit = $0.01), metered by token consumption at each model's rate. Replaced the legacy 300 premium requests/month on June 1, 2026; extra credits purchasable.	✓ Jul 29, 2026	Official page
Model access	Unlimited base-model chat; frontier models consume AI Credits	✓ Jul 29, 2026	Official page
Code completions	Unlimited	✓ Jul 29, 2026	Official page

FAQ

Concrete answers from the tracked data above.

Which AI plan has the highest usage limits for $20/month?

By raw message volume, ChatGPT Plus (160 flagship messages per 3 hours) allows more messages than Claude Pro (~45 messages per 5 hours). Google AI Pro gives 4× Gemini's standard compute allowance, and Perplexity Pro allows 300+ Pro searches per day. Which cap you actually feel depends on your workflow.

What happens when I hit a limit?

ChatGPT falls back to a lighter model until the rolling window resets. Gemini's allowance refreshes every 5 hours inside a weekly cap. Claude stops until the 5-hour window (or weekly cap) resets. Perplexity continues on unlimited quick searches. Cursor continues on on-demand billing, and Copilot draws on purchasable AI Credits.

Do providers publish all of these limits?

No. Some caps (like Claude's free tier or ChatGPT's image limits) are intentionally unpublished and vary with demand. Where a provider publishes no number, this table says 'Not published' rather than inventing one.

How often is this data checked?

Each fact carries its own verification date — the table shows exactly when every number was last checked (latest: Jul 29, 2026). Changes are recorded in the changelog with before/after values.

ChatGPT usage limits Claude usage limits Gemini usage limits Limit changelog Compare AI plans

All plans at a glance

Recent limit changes

ChatGPT Free limits

ChatGPT Plus limits

ChatGPT Pro limits

Claude Free limits

Claude Pro limits

Claude Max limits

Gemini Free limits

Google AI Pro / Gemini Advanced limits

Google AI Ultra limits

Perplexity Free limits

Perplexity Pro limits

Cursor Hobby / Free limits

Cursor Pro limits

GitHub Copilot Free limits

GitHub Copilot Pro limits

FAQ

Related pages