AI agent comparison (2026)
Manus, Operator, Claude Computer Use, CloudyBot—by pillar.
Pricing
AI that does the work—
AI that does the work—
at a price you can predict.
Every plan has clear limits. When you hit your cap, the service pauses—it doesn’t charge you more. Why predictable pricing →
When checkout includes a trial (often 7 days by default), Stripe may mark your subscription as trialing with $0 until the trial ends. Limits during trialing follow your workspace configuration (commonly Pro-level); then your chosen paid plan or Free if you cancel.
Runtime plan keys are free, growth, pro, and max. Legacy names normalize to current tiers: starter/base → Growth, agency → Max.
Hard caps — no overages
Cancel anytime
Free tier — no card required
Free
$0
Try before you buy. No credit card. Forever free.
- 30 AI Tasks / month
- 10 cloud browser minutes / month
- 5 web searches / month
- 50 MB storage, 10 files
- Tier-1 models only (DeepSeek, Haiku, GPT-4o mini)
- 7-day chat history
- Parallel helpers — not included
- Dashboard only (no WhatsApp)
Forever free · Hard caps on included use
Most Popular
Growth
$19/mo
Best value for daily use.
- 1,500 AI Tasks / month
- 750 cloud browser minutes / month
- 150 web searches / month
- 1 GB storage, 100 files
- Tier-1 + Tier-2 + Tier-3 (Claude Sonnet)
- 90-day chat history
- Up to 5 parallel helpers
- 100 WhatsApp proactive sends
Hard monthly caps · Cancel anytime · No surprise meter overages
Pro
$39/mo
Power users. All models.
- 3,000 AI Tasks / month
- 1,500 cloud browser minutes / month
- 300 web searches / month
- 2 GB storage, 200 files
- All models (Claude Opus, GPT-5.4)
- 180-day chat history
- 200 WhatsApp proactive sends
- Up to 10 parallel helpers
Hard monthly caps · Cancel anytime · No surprise meter overages
Max
$149/mo
Heavy automation. Unlimited history.
- 7,000 AI Tasks / month
- 3,500 cloud browser minutes / month
- 700 web searches / month
- 5 GB storage, 500 files
- All models + priority queue
- Unlimited chat history
- Up to 20 parallel helpers
- 500 WhatsApp proactive sends
Hard monthly caps · Cancel anytime · No surprise meter overages
Enterprise
Custom
Volume, security review, SSO, and bespoke integrations.
- Custom limits & contracts
- Dedicated onboarding
- Not available in self-serve checkout
Custom limits · We’ll scope your rollout
Why upgrade to a paid plan?
Self-serve paid tiers (Growth, Pro, Max) raise the same hard gates the product enforces. The table below matches the current runtime plan limits (Free → Max).
| What goes up when you upgrade | Free | Growth | Pro | Max |
|---|---|---|---|---|
| More AI Tasks per month | 30 | 1,500 | 3,000 | 7,000 |
| More cloud browser minutes / month | 10 | 750 | 1,500 | 3,500 |
| Higher model tier cap (unlocks more capable models in the picker) | 1 | 3 | 4 | 4 |
| More scheduled jobs (cron) | 0 | 6 | 15 | 50 |
| More specialist slots (employee slots) | 0 | 3 | 5 | 999 |
| More web searches / month | 5 | 150 | 300 | 700 |
| More VPN proxy session blocks / month (5-minute blocks) | 0 | 25 | 50 | 150 |
Enterprise is contract-based; limits are agreed separately. The numbers above are the same monthly caps used for enforcement on Free through Max (AI tasks, browser minutes, model tier, scheduled jobs, specialist slots, web searches, and VPN proxy blocks).
How limits work
- Hard caps: Each plan lists monthly limits (AI Tasks, browser minutes, web searches, and more). When a meter hits zero, that feature stops—CloudyBot does not bill automatic pay-per-use overages on top of those caps.
- AI Tasks + wallet: Your plan includes a monthly AI Task allowance that resets each billing cycle. Optional task top-ups (and referral bonuses when enabled) add the same task units to a credit wallet. Before each chat turn, CloudyBot checks plan tasks plus wallet balance; new chat is refused only when both are insufficient, with a clear error (for example: “Tasks exhausted. Buy credits or wait for next billing cycle.”).
- Trials: When Stripe checkout includes a trial (often 7 days by default), your subscription can be trialing. Limits during trialing follow server configuration (commonly Pro-level plan limits; operators can configure otherwise).
Complex models cost more tasks per message (e.g. Claude Sonnet = 6×, Claude Opus = 15×). Standard models = 1×. Task cost is shown next to each model in the picker.
Running low mid-month? Top up your task wallet — extra AI Tasks available at any time. They never expire.
Understanding AI Tasks & limits
What is an AI Task?
1 Task ≈ 1 message on standard models (1×). Larger tasks (file analysis, browser automation) may use 2–5 tasks. Complex models use multipliers: Kimi/GPT-4o = 3×, Claude Sonnet = 6×, Claude Opus/GPT-5.4 = 15×. The model picker shows the cost per message.
Compare
Full Plan Comparison
See exactly what's included in each tier.
| Feature | Free | Growth ⭐ | Pro | Max | Enterprise |
|---|---|---|---|---|---|
| Price | $0 | $19/mo | $39/mo | $149/mo | Custom |
| AI Tasks | |||||
| AI Tasks / month | 30 | 1,500 | 3,000 | 7,000 | Custom |
| Parallel helpers (sub-agents) | 0 | 5 | 10 | 20 | Custom |
| Scheduled cron jobs | 0 | 6 | 15 | 50 | Custom |
| Employee slots (Specialists) | 0 | 3 | 5 | Unlimited | Custom |
| Requests / hour (abuse cap) | 10 | 80 | 120 | 200 | Custom |
| Chat history | 7 days | 90 days | 180 days | Unlimited | Custom |
| Rolling memory | ✓ | ✓ | ✓ | ✓ | ✓ |
| Moderation | ✓ | ✓ | ✓ | ✓ | ✓ |
| Cloud Browser | |||||
| Cloud browser minutes / month | 10 | 750 | 1,500 | 3,500 | Custom |
| Billed in minutes while the browser runs. Proxy add-on uses separate 5-minute blocks (see below). | |||||
| Live view | — | ✓ | ✓ | ✓ | ✓ |
| File Workspace | |||||
| Max files | 10 | 100 | 200 | 500 | Custom |
| Storage | 50 MB | 1 GB | 2 GB | 5 GB | Custom |
| Git repo storage (connected repos) | 100 MB | 2 GB | 5 GB | 20 GB | Custom |
| Usage meters | |||||
| Web searches / month | 5 | 150 | 300 | 700 | Custom |
| Code exec runs / month (sandbox) | 0 | 50 | 200 | 1,000 | Custom |
| AI images / month | 0 | 10 | 30 | 60 | Custom |
| AI videos / month (max 10s each) | 0 | 1 | 3 | 6 | Custom |
| Social posts / month (Zernio) | 0 | 30 | 60 | 150 | Custom |
| Connected social accounts | 0 | 5 | 10 | Unlimited | Custom |
| Channels | |||||
| Dashboard chat | ✓ | ✓ | ✓ | ✓ | ✓ |
| — | ✓ | ✓ | ✓ | ✓ | |
| WhatsApp proactive sends | 0 | 100 | 200 | 500 | Custom |
| Proxy sessions / month | 0 | 25 | 50 | 150 | Custom |
| Models | |||||
| Tier-1 (DeepSeek, Haiku, GPT-4o mini) | ✓ | ✓ | ✓ | ✓ | ✓ |
| Tier-2 (Kimi, GPT-4o) | — | ✓ | ✓ | ✓ | ✓ |
| Tier-3 (Claude Sonnet) | — | ✓ | ✓ | ✓ | ✓ |
| Tier-4 (Claude Opus, GPT-5.4) | — | — | ✓ | ✓ | ✓ |
| Priority API queue | — | — | — | ✓ | Custom |
| Support | |||||
| Support level | Community | Priority | Priority | Dedicated | |
How Hard Caps Work
AI Task Cap
When you reach your monthly AI Task limit, chat pauses until the next billing cycle. Complex models cost more tasks per message (multipliers shown in model picker). Top up mid-month with your task wallet.
Browser Minutes Cap
Browser runs on-demand — starts when needed, stops automatically. If you hit your browser minute cap, browser requests pause until next cycle. Resets monthly.
Storage Cap
Uploads pause when you hit file count or storage limit. Delete old files or upgrade to add more.
Rate Limit
Requests per hour prevent abuse. Wait a few minutes if you hit the limit, or upgrade for higher throughput.
Compare & Evaluate
Not sure which plan or approach fits?
Use these guides to compare hosted vs self-hosted, understand agent architecture, and evaluate predictable AI pricing.
Hosted vs self-hosted AI
Full cost and operational tradeoff breakdown.
AI agent vs chatbot
Choose the right workflow model for your team.
Hard caps vs pay-per-use
Why predictable spending controls matter for adoption.
FAQ
Pricing Questions
How does CloudyBot compare to Manus AI or OpenAI Operator?
Both are powerful task-oriented agents. Manus is typically usage-based without the same hard monthly ceiling model. ChatGPT Operator runs inside ChatGPT with bundled messages—great for interactive work, but not a substitute for cron Specialists, Workflow Architect deployments, or a standalone PWA with push. CloudyBot adds scheduled duties, hard AI Task + browser caps (service pauses instead of overages), WhatsApp/Telegram on paid tiers, and surgical file edits that stretch each task. See AI agent comparison (2026) on this site.
How many Specialists, cron jobs, and “employees” do I get per plan?
Limits are enforced per subscription tier (see the plan cards above). In the product, Specialists map to employee slots; cron jobs cap how many scheduled duties you can run; higher tiers add more browser time, AI Tasks, social posts, and integrations. Confirm limits on this page before checkout.
What is an AI Task?
1 AI Task is about one message on standard models (1×). Larger tasks like file analysis or browser automation may use 2–5 tasks. Complex models use multipliers: Kimi/GPT-4o = 3×, Claude Sonnet = 6×, Claude Opus/GPT-5.4 = 15×. The model picker shows the cost per message. Your usage is hard-capped so you are not billed uncapped post-hoc token overages, and you can top up mid-month with your task wallet when you purchase credits. Surgical file edits (line-level patches instead of rewriting whole files) help large documents use fewer tasks.
What happens when I hit my AI Task limit?
Each billing period includes a fixed monthly AI Task allowance for your plan (shown on this page). Before a chat turn starts, CloudyBot checks remaining plan tasks plus your optional credit wallet—both use the same task units. If the combined total is zero, new chat requests are rejected with HTTP 429 and a clear error (for example: Tasks exhausted. Buy credits or wait for next billing cycle.) until you buy a task top-up, receive wallet credits (for example referral bonuses when enabled), or your plan tasks reset at the next billing cycle. After monthly plan tasks are used up, further turns draw from the wallet until it is empty.
Do you charge token overages automatically?
No. AI usage is billed as your subscription plus optional one-time task packs you choose to buy. There is no automatic unlimited pay-per-token overrun that bills beyond your plan without an explicit credit purchase. Extra chat beyond the monthly plan allowance uses prepaid wallet task units when you have a balance.
How do credit top-ups work?
You can buy a one-time Stripe credit pack from billing when your operator configures a credit price. The purchase adds task units to your credit wallet. Default pack size is 100 tasks per purchase unless your deployment configures a different amount. The wallet is used after your monthly plan task allowance for the period is exhausted. Referral bonuses, when enabled, credit the same wallet.
Is there really a free tier with no credit card?
Yes. Sign up with just your email, start chatting immediately. 30 AI Tasks is enough to thoroughly evaluate the product. No credit card required unless you choose to upgrade. The free tier is genuinely free, forever.
What happens when I hit a cap?
The service pauses for that meter—no surprise automatic overage charges. AI Tasks: new chat is blocked only when both your monthly plan tasks and your credit wallet are at zero; until then, wallet tasks can still be used after the monthly bucket is used up. Browser minutes, storage, WhatsApp sends, and other meters pause their respective features when exhausted. You can upgrade or buy explicit top-ups where available, or wait for the monthly reset on plan-bucket limits.
What counts as a cloud browser minute?
Browser minutes are on-demand. The browser starts automatically when you need it and stops when you are done—you are only charged for the time it is actively running. No need to manually disconnect to save minutes. Minutes reset at the start of each billing cycle.
Can I upgrade or downgrade mid-cycle?
Yes, within Stripe rules. Use the billing controls in your dashboard to open the Stripe Customer Portal or start checkout for a different paid tier. Stripe handles invoicing, cancellation, and any proration according to your subscription and Stripe settings. CloudyBot applies plan limits from your subscription record after Stripe updates it (usually within a short delay while webhooks process).
Do you offer team or enterprise plans?
Not yet. CloudyBot is currently a per-user product. Team billing with shared workspaces is on the roadmap. For now, each team member needs their own account. Check our roadmap for updates.
What's your refund policy?
Because our free tier lets you thoroughly evaluate the product before paying, we generally don't offer refunds. However, if you experience a technical issue that prevents usage, contact us and we'll make it right.