- 05
- June
The most direct question for anyone buying Claude AI is: "How many messages does each plan give me?" The quick, decision-ready answer: Free is very limited · Pro ($20/mo) ≈ 45 messages every 5 hours (about 200/day) · Max 5× ($100) ≈ 225 messages / 5 hours · Max 20× ($200) ≈ 900 messages / 5 hours · while Team/Enterprise are team plans with admin-managed quotas. This article explains "what a limit is, how it's counted, and exactly how much each plan gives" — in concrete numbers.
What is a limit? (in 4 points)
- A limit = the number of "messages" you can send before you hit the cap and must wait for a reset
- Counted in 5-hour windows — use it up, then wait until the 5 hours pass for the quota to refill
- Plus a weekly cap — to prevent burning a whole week's usage at once
- Length matters — long messages, big file attachments, or long conversations consume more than one short message
How many messages per plan — the real numbers
This is the table you can actually decide from (approximate figures from real-world usage and sources — see the note below).
| Plan | Messages / 5 hrs (approx.) | Per day (approx.) | Price | Best for |
|---|---|---|---|---|
| Free | Few (drops at peak) | Limited | $0 | Trying it out |
| Pro | ~45 | ~200 | $20/mo | Serious personal use |
| Max 5× | ~225 | ~1,000+ | $100/mo | Heavy use / coding |
| Max 20× | ~900 | ~4,000+ | $200/mo | All-day, rarely hits limits |
| Team | More than Pro per seat (Premium far more than Standard) | Per seat | $20–100/seat/mo | Teams of 5–150 + admin |
| Enterprise | Seat + metered API usage | No session-style limit | custom | Large orgs / very heavy use |
Note: these are approximate — Anthropic doesn't publish fixed numbers and adjusts them (in May 2026 it raised Claude Code limits) · actual counts depend on message length, attachments, conversation length, and model · "per day" assumes usage spread across the day · prices are USD before 7% VAT.
What is a "message," how long is it — and why some days run out fast
The most important question before trusting the "45 messages" figure is: how is a "message" counted?
- "One message" = one time you type and hit send (one prompt/turn) — not a session
- "Session" = the 5-hour window (starting from your first message), within which you can send ~45 messages (Pro)
- There's no fixed length per message — Claude actually counts "tokens" (amount of data), not headcount of messages. So "45" is just an average for typical-length messages
Because it's token-based, the real count depends on how "heavy" each message is:
- Short messages → more than 45 · long messages / big attachments → fewer (one big file = several short messages)
- The longer the chat, the faster it compounds — each reply re-reads the whole history; e.g. a 50K-token history + your new question = 50K+ every turn → start a new chat when you change topics to save a lot
- Stronger models (e.g. Opus) = consume more than smaller models
- Counts across all surfaces — claude.ai + Claude Code + Desktop share one quota
- Peak hours = Free/Pro may get a smaller effective quota; higher plans get priority
Quota-saving tips: start a new conversation when you change topics (don't keep one chat endless), attach only necessary files, and use smaller models for easy tasks — this makes Pro's ~45 messages go much further.
Context window — how much it can read at once
Another limit people confuse: the "context window" = how much data Claude can read/remember in a single conversation (different from message count).
- 200K tokens (standard on all plans) ≈ reading a very long document at once (roughly a thick book / a small whole codebase)
- It can be larger — Enterprise reaches 500K on some models, and up to 1M tokens on Opus 4.6 in Claude Code (Max / Team / Enterprise, since March 2026)
For organizations — limits that matter more than message counts
When buying for a whole team, what decides which plan you need usually isn't "message count" but governance.
| What you need | Available from |
|---|---|
| Data not used to train models (default) | Team / Enterprise / API |
| Admin seat management + analytics | Team and up |
| SSO + audit + data residency choice | Enterprise |
Read on: Team vs Enterprise — which to choose · Data Governance & Security.
Summary — choose a plan by usage volume
| If you... | Choose |
|---|---|
| Ask a few dozen times a day, general work | Pro ($20) |
| Heavy use / coding / hit Pro's limit often | Max 5× or 20× |
| Team of 5–150 wanting admin + one bill | Team |
| Mandate SSO/audit or very heavy org-wide use | Enterprise |
The easiest number to remember is "Pro is about 45 messages every 5 hours," and Max is 5× and 20× of that. If you hit Pro's limit often, it's time to move up to Max; teams with many users move to Team/Enterprise so admins can manage it.
- A guide to choosing a Claude plan for Thai organizations
References
- Claude Help Center — How do usage and length limits work?
- Claude Help Center — What is the Pro plan?
- Claude — Plans & Pricing
Not sure which plan is enough for your team?
Grand Linux assesses your usage volume and team size, then issues a Claude Team/Enterprise quote in baht with a Thai tax invoice.
Get advice / request a quoteTel 02-347-7730 | sale@grandlinux.com
