One-line positioning
A general AI agent — you give a goal, it plans, drives a browser, writes code, and delivers, not a chat box
Manus is a general AI agent, iterated to the 1.6 series in 2026 (1.6 / 1.6 Lite / 1.6 Max). You give a goal in plain language; it breaks steps, drives browser and code, connects Google Drive/Slack/email, runs asynchronously in the cloud, and notifies you when done. It ships research reports, full sites/apps, slides, and image/video. Pricing is "monthly fee + credits": Free $0 (300 credits/day, 1 concurrent), Pro $20/mo (4,000 credits, 20 concurrent), Pro $40/mo (8,000 credits, 7-day trial), up to $20,000/mo; Team from $20/seat/mo; annual saves 17%.
Getting Started (based on community-tested notes, not first-person)
Note: compiled from official docs and user testing, rewritten and condensed by editors — not a claim that we tested each point ourselves.
- Start free: Free gives 300 credits/day and 1 concurrent task, enough for basic research and slides; a temporary free campaign before 2026-08-25 loosens limits.
- Upgrade to Pro: for regular competitive analysis / content research / site building, pick $20 or $40/mo — 4,000–8,000 credits + 20 concurrent + site deploy + Wide Research.
- Teams use Team: collaboration, permission control, opt-out of data training, SSO, from $20/seat/mo with a shared credit pool.
Real Pain (community consensus)
Note: recurring pain points reported across the community, condensed by editors, not individually verified by us.
- Credits are not unlimited: the fee buys credits, not unlimited tasks; complex long tasks burn fast, heavy users must keep upgrading.
- Free is cut down: the free tier only runs 1.6 Lite; complex tasks lose quality, Max needs payment.
- Peak-hour queue: free tasks queue visibly; paid tiers get priority, urgent work needs an upgrade for timing.
- Fragmented price ladder: the public page shows only a few starting points; real tiers run $20 to $20,000, pick inside the pricing tool.
- Pre-08-25 free campaign: during it Free/paid can run 1.6/1.6 Lite without credits, but Max stays excluded, then reverts.
Hard Comparison (one-line verdicts)
- WorkBuddy: Tencent's desktop Agent drives your local files and remote-controls across devices; Manus wins on cloud-async, site deploy, and cross-app (Slack/email).
- Trae Work: ByteDance evolved from coding, dev-friendly; Manus wins on general-task breadth (research/slides/video alike).
- Qwen Office / Doubao Pro: domestic Agents do office automation; Manus wins on global general ability and English ecosystem, but weaker localization.
Who It's For
- For: individuals and small teams doing research/site/slides, "drop goal, wait for result" people, cross-app automation needs.
- Not for: those wanting only light chat, budget-tight high-frequency long-task users (credits burn fast), or deep local-file operators (pick WorkBuddy).
Editor's Take
Note: our view from a "deliver results" angle, not "smart chat" angle.
Manus's best fit is "intern" — you give a goal, it plans, uses tools, and delivers, while you can close the laptop. For non-coders who need a site or research, this "drop goal, wait for result" experience is a dimension drop. But the cold water: the credit system is a hidden cost, long tasks burn faster than you think, and free-tier 1.6 Lite shows its limits on complex work. Our advice: run one real task on Free first (not just the demo), confirm it truly delivers, then upgrade to Pro; teams use Team but check the data-training opt-out before committing; don't park urgent work in the free queue. Don't be fooled by "general agent" into thinking it does everything — it's strong on general flow, but deep specialty still needs a human in the loop.
Ratings
Autonomous planning and delivery: ★★★★★
General-task breadth: ★★★★★
Ease of use (drop goal, get result): ★★★★★
Value (credit-system caveat): ★★★★☆