Kimi
“The AI that reads long documents whole” — a reviewer who reads the entire stack of papers and comes back with the answer
What is Kimi?
Kimi is a conversational AI from Moonshot AI in Beijing, China. Its selling points are reading very long documents in one pass and agent features that carry out multiple steps on their own.
- 1Its core use is reading long documents in one go
- 2The official guide lists paid memberships only
- 3Under official policy, data is stored inside China
What sets it apart
These are the differences we found by running the same material through several tools in our training sessions.
Reads long text without breaking it up
You no longer have to chop a document into pieces and re-explain the context each time. The official membership guide lists million-token conversations as a perk of the higher tiers, so uploading a long document once and asking question after question is the basic way to use this tool.
It works through multiple steps on its own
The official membership guide lists the number of agent runs as the criterion that separates the plans. That means once you give an instruction, it carries on through searching, organizing, and producing the deliverable, so it works more like an assistant you delegate to than a conversational chatbot.
Pricing is tied to a single metric
The official guide says all plans share a single credit pool. Deployment, research, coding, and work features do not run on separate budgets but draw from one pool, which makes it relatively simple to calculate monthly spend.
⚖️ The weaknesses are clear, too
The official privacy policy states that personal information is stored within the territory of the People’s Republic of China, and that inputs and outputs are used to improve the model. Opting out of training use is described as a separate request to customer support rather than a single setting, which makes the process too cumbersome for company materials.
💡 In plain EnglishThink of a law-firm reviewer who reads the whole stack of papers and comes back to point out only the key issues. Its strength is cross-checking, not summarizing.
The biggest time AI saves at work is on “reading long documents.” Seeing how far a tool aimed squarely at that one task has come helps you judge what your organization can hand off to it.
Same job, different approach
You need to read a 100-page public tender notice and decide whether we meet its requirements
The old way
- Reads from start to finish, highlighter in hand
- Copy the eligibility requirements and scoring table out separately
- Checks them against our company’s conditions one by one
- Rereads from the top in case something was missed
With Kimi
- Upload the entire notice file at once
- Ask for only eligibility, scoring, and deadlines in a table
- Describe our requirements and have it check them against the document
- A person checks only the flagged items against the original
Common uses
Reading public notices and regulations
Upload a long public document, such as a bid notice or a government announcement, and pull out just the eligibility rules and deadlines as a table. Always check the original text.
Cross-Check Issues in Long Reports
Upload several reports together and have it find the points where they conflict. A person judges the conflicts.
Hand off an entire research task
Give the agent a topic and the shape of the deliverable, then let it work. Credits drain quickly, so start small.
Who it’s for
Best for
- ✓Planning and general affairs managers who often need to read long public documents
- ✓People who often cross-check multiple sources to pull out key issues
- ✓Training managers who want to try agent features firsthand
Better served by other tools
- –Work that involves internal documents or contracts
- –Organizations that need a fixed monthly cost
- –People looking to borrow Korean customer-service phrasing
Can we use it at work?
The questions that adoption leads and security teams ask first.
Verification note: We confirmed the operating entity is Moonshot AI (北京月之暗面科技), based in Beijing. However, we could not find in the official documents how overseas data-transfer notices are handled for users in Korea, or whether a separate region is offered for enterprises. Legal review is needed before company-wide adoption.
10-minute quick start
- 1
Go to kimi.com, sign up, and ask one short question in Korean.
- 2
Upload one long public document and ask for only the eligibility requirements and deadlines in a table.
- 3
If the results are useful, choose the tier you need from the official membership page.
- 4
Start agent features with small tasks and measure how many credits they use first.
Copy-and-paste prompts
Replace only the [bracketed] parts, then paste it in.
From the uploaded document, pull out only ① eligibility requirements ② scoring criteria ③ submission deadline and method, and put them in a table. For any item not in the document, write “Not in document,” and mark which page of the original each row comes from.
Here are our company’s conditions: [list conditions]. Check them one by one against the requirements in the uploaded document and sort them into three groups: ① met ② not met ③ cannot determine. For each unmet item, add one line on what we would need to meet it.
Read all the documents I uploaded together and find only the passages that cover the same issue but say different things. Show the document name and the relevant sentence side by side, and mark which source is more recent.
Research [topic] for me. The deliverable is one table and a ten-line summary. Before you start, show me your plan and the estimated credits, and begin only when I reply 'go ahead'.
From now on, answer using only what is in the documents I uploaded. Do not fill gaps with knowledge from outside the documents; if you do not know, just write 'Not in the document'. Keep following this rule until I say to lift it.
Using it with other tools
Linking two or three tools changes your work more than any single tool does.
Reviewing a public notice (two hours)
Sorting a pile of materials (half a day)
Before you start
- Do not upload internal documents or contracts. Under the official policy, data is stored within mainland China.
- Opting out of training use is described as a request to customer support, not a setting. Check the procedure first.
- Credits come from a single pool shared within each plan. Running agents uses them up faster than you might expect.
- Reading a long document does not make an answer correct. Have a person check the page numbers it cites.
Pricing: personal use and work use
Plans that individuals pay for are different from plans that organizations contract for by member.
For individual use
About 30 agent runs, 20 projects and 20GB of storage
About 60 agent runs, 2 concurrent tasks
About 150 agent runs, goal mode available
About 360 agent runs, or conversations in the millions of tokens
Using it at work
Output is $15.00 per 1M tokens
Output is $4.00 per 1M tokens
Checked 21 Sep 2026 · Prices are as listed on Kimi’s official help page for membership pricing and in the API price list in its official developer docs. Memberships are priced in Chinese yuan (¥) and the API in US dollars. The currency you are actually billed in may vary by your country and payment method. Prices change without notice — check the official pages before you pay. Kimi official membership pricing ↗ · Kimi official membership overview ↗ · Kimi official API pricing ↗ · Kimi official privacy policy ↗
Where Kimi falls short
We don’t just list the positives. Here is what fell short when we used it.
- It asks for payment first
You want to try it out before deciding, but the official pages list only paid memberships. Having to enter a card before you know how useful it is gives you pause.
- Credits are hard to track
Because credits are pooled into a single balance for each plan, it is hard to tell how much a single task you just ran cost. You only realize you have used a lot at the end of the month.
- The settings screens are not in Korean
Key screens, such as the opt-out from training use and the payment terms, appear only in Chinese and English. You end up clicking with a translator open, then double-checking in case you clicked the wrong thing.
Tools to compare
FAQ
Last updated 22 Sep 2026 · The ratings on this page come from John Kim, CEO of Nanum Consulting, who has used these tools firsthand in training and real work.
Put Kimi to work in your organization
Send us a message and we’ll reply within 24 hours. We’ll propose a curriculum tailored to your industry and job roles.