AI Chat · Anthropic
✓ Verified Jun 2026
Last audited: 2026-06-29
Claude:
What the pricing page doesn't tell you
Based on public docs, community reports, and third-party benchmarks — not legal advice or official statements.
Hidden Behaviors
documented & verified
Transparency
52%
of key policies disclosed upfront
Last Verified
Jun 2026
$20/mo · Claude Pro
Hidden Behaviors
5 documented
Behaviors confirmed through user reports, help docs, and independent testing — none disclosed on the pricing page.
Dual cap system — two separate limits, either can block you
▾
Official
Not clearly disclosed on pricing pageThe pricing page describes Pro as having 'usage limits' but does not explain the 5-hour AND weekly cap structure.
Observed (community)
Both caps operate independentlyHitting the weekly cap locks you out completely even if your 5-hour session bar shows remaining headroom.
Estimated (not verified)
~45 msgs per 5h windowBased on community reports — Anthropic publishes multipliers (Pro=1x, Max=5x), not absolute numbers.
Weekly reset is account-specific — not Monday midnight
▾
Official
Not documented publiclyAnthropic does not publish the weekly reset schedule or explain that it varies per account.
Observed (community)
Reset tied to account creation dateCommunity testing suggests weekly caps reset on a fixed day/time assigned at account creation — not a universal Monday reset.
Estimated (not verified)
Many guides are wrongWidespread guides claiming 'resets on Monday' appear to be inaccurate for most accounts.
Current 50% limit boost is a promo — expires July 13, 2026
▾
Official
Disclosed on Anthropic blogMay 13, 2026 post confirms the 50% weekly limit increase is promotional, ending July 13, 2026.
Observed (community)
Not surfaced in-appNo in-product banner or countdown warns users the current higher limits are temporary.
Estimated (not verified)
Limits revert after Jul 13After expiry, weekly caps return to pre-May 2026 baselines — expect a noticeable reduction for heavy users.
Claude.ai, desktop app, and Claude Code share the same quota pool
▾
Official
Not disclosed on pricing pageNo mention of shared quota across clients in plan comparison or FAQ.
Observed (community)
Coding drains chat quotaCommunity reports confirm heavy Claude Code sessions deplete the same usage pool as claude.ai chat.
Estimated (not verified)
Power users hit limits fasterDevs switching between Code and chat may exhaust limits ~2× faster than chat-only users.
Progress bars shown, but exact token counts never published
▾
Official
Usage bars exist in SettingsSettings → Usage shows two progress bars for session and weekly limits — better than ChatGPT.
Observed (community)
No absolute numbers shownBars show percentage consumed, not tokens or message counts. Anthropic publishes only relative multipliers.
Estimated (not verified)
Planning usage is hardWithout knowing the absolute cap, users cannot estimate how many messages remain before hitting a wall.
Real Limits
Reset Cycle
Rolling 5-hour window + separate weekly cap (resets on account-assigned day, not Monday)
Stated Cap
~45 msgs / 5h (reported) — Anthropic publishes multipliers only, not absolute counts
Usage Meter
Yes — Settings → Usage shows two progress bars, but no exact token figures
Speed
Claude Haiku 4.5: ~500–650ms time-to-first-token (benchmark range, third-party)
Privacy Default
Does NOT train on Pro data by default — opt-in only (stronger than ChatGPT)
⚠ Dual-cap trap
Your weekly bar can hit 100% while your 5h session bar still shows headroom — locking you out until your account's weekly reset, which may not fall on Monday.
Policy History
Confirmed changes at the same price — documented with source type and dates.
Mar 28, 2026
Peak-hour throttling silently added (5–11 AM PT weekdays) — acknowledged by Anthropic on Mar 31
OBSERVEDcommunity-reported; Anthropic acknowledged March 31, 2026
May 6, 2026
5-hour limits permanently doubled for Pro, Max, Team — peak-hour throttling removed
OFFICIALanthropic.com/news/claude-usage-limits-update
May 13, 2026
Weekly limits raised 50% as a promotion — expires July 13, 2026
OFFICIALanthropic.com/news/claude-usage-limits-update
Who It's For
Same data, different verdict depending on how you use it.
✓ Works well for
Long-document analysis
200k+ context window — handles full codebases or book-length docs.
Writing quality
Consistently rated best prose quality among mainstream AI assistants.
Privacy-first users
Does NOT train on Pro data by default — opt-in only.
✗ Problematic for
Burst chat sessions
Dual 5h + weekly cap — hitting either locks you out completely.
Image/video generation
No image or video generation on any Claude plan.
Predictable limits
Limits changed 3 times in 2026 alone — current 50% boost expires Jul 13.
Transparency Breakdown
How much is disclosed about each dimension — before you pay.
Real Headroom
~45 msgs/5h reported — no official absolute count published. Dual cap complicates planning.
multipliers only, not absolute numbers
Response Speed
Haiku 4.5: ~500–650ms TTFT range — among the fastest mainstream models tested.
third-party benchmark only
Limit Transparency
Usage bars exist in-app — better than most. Exact token counts and weekly reset day not disclosed.
bars visible, counts not
Data & Privacy
Pro data not used for training by default — opt-in only. Strongest default privacy posture in category.
best disclosed dimension
Reliability
~2 incidents/30d — ~99% uptime (est.). Status page is public.
status page public
Overall Transparency Score ⓘ HOW?
Weighted estimate — see methodology for formula
52%
Compare Transparency
Frequently Asked
Why did I get blocked even though my session bar wasn't full?
Claude has two separate caps: a 5-hour rolling window and a weekly cap. Hitting the weekly cap locks you out even with session headroom remaining.
Can I see my exact usage on Claude?
Settings → Usage shows progress bars for both limits, but Anthropic never publishes exact token numbers — only relative multipliers (Pro = 1x, Max = 5x or 20x).
Is Claude's current 50% higher weekly limit permanent?
No. It's a promotion running through July 13, 2026. Expect limits to revert to standard baselines after that date.
Do Claude Code and claude.ai share the same usage pool?
Based on community reports, yes — Claude Code sessions draw from the same quota as claude.ai chat, which can exhaust limits faster for developers.
Get alerted when policies change
We track silent limit cuts before most users notice them.