Upstage Solar
“A Korean-built engine that reads company documents” — an archivist that turns stacks of paperwork into tables
What is Upstage Solar?
Upstage Solar bundles AI models and document-processing features built by a Korean company. It converts documents such as contracts and quotes into machine-readable form so you can connect them to internal systems.
- 1You buy and use a Korean-built model directly
- 2Documents are read and billed per page
- 3It’s not a tool you use alone — it’s a component you plug in
What sets it apart
These are the differences we found by running the same material through several tools in our training sessions.
Reading documents is its core business
There is a dedicated feature that converts documents mixing tables and stamps, such as contracts, quotes, and transaction statements, into a machine-readable structure. Pricing is per page, so you can calculate how much your company would spend each month, which makes a big difference for a manager who has to put together a quote.
You contract with a Korean company
Both the company that built the model and the one that writes the contract are Korean. For organizations where sending data to overseas services stalls adoption, this means one more option worth considering. According to official guidance, it also offers an option to run it inside your own company rather than only in the cloud.
Try it first, no card needed
The official pricing page says each document-processing agent comes with ten free runs, so you can test it on real documents without registering a card. You can first check whether it reads your company’s paperwork correctly, then talk about cost.
⚖️ The weaknesses are clear, too
This isn’t a tool you open and use on your own. Pricing is per token and per page, and the results only pay off when connected to other systems, so a developer or IT systems manager has to get involved. That’s different from a training manager trying a tool alone this afternoon and feeling the benefit right away.
💡 In plain EnglishThink of an archivist guarding the company’s document vault. It doesn’t write reports itself, but it finds the paperwork, organizes it into tables, and hands it over.
The point where Korean companies most often get stuck on AI adoption is not performance but where their data gets sent. The option of contracting with a Korean company and even hosting it in-house moves that discussion one step forward.
Same job, different approach
Staff retype hundreds of invoices by hand every month
The old way
- A staffer keys in numbers while glancing back and forth between paper and screen
- Table layouts differ by vendor, so you can’t set rules
- One typo means redoing everything at month-end close
- When someone leaves, their method leaves with them
With Upstage Solar
- Feed in a document and it reads it with the table structure intact
- Define the fields to extract in advance and fill them in automatically
- People only check the few pages it could not read
- The method stays in the system and keeps working even when the person in charge changes
Common uses
Reading contracts and invoices
It pulls amounts, periods, and items from paperwork whose format differs by vendor and turns them into a table. Once retyping disappears, people are left with just the checking.
Add internal document search
Load your regulations and past proposals so employees can find answers by asking in plain language. Start by deciding whether your data will stay inside or outside the company.
Check how a Korean model performs
Add one more option to a meeting that has only looked at overseas models. Run the same test on your own business documents, and the meeting gets shorter.
Who it’s for
Best for
- ✓Departments that process documents in bulk
- ✓Organizations that find it hard to send data overseas
- ✓Companies with someone who can connect AI to internal systems
Better served by other tools
- –People looking for a tool to handle their writing for them
- –People who want to use it right away without developers
- –People who want a single flat monthly fee
Can we use it at work?
The questions that adoption leads and security teams ask first.
Please note: we could not confirm on the official pricing page whether input documents are used for training, how long they are retained, or whether data is stored in Korea. If you are considering adoption, confirm these in writing before signing.
10-minute quick start
- 1
Create a console account on the official site, preferably under a company-owned account.
- 2
Pick ten sample documents with personal data removed and run them through Document Parse.
- 3
A person checks the amounts, dates, and items in the extracted results against the originals.
- 4
Multiply your monthly volume by the per-page rate to estimate the cost, then decide whether to adopt it.
Copy-and-paste prompts
Replace only the [bracketed] parts, then paste it in.
From the document below, extract only the following fields into a table: [counterparty], [contract period], [total amount], [payment terms], [termination clause]. Leave any field that is not in the document blank and never guess. [Paste document content]
I want to measure document recognition accuracy using 50 of our company’s [transaction statements]. Help me design the test: which fields to check when scoring each document as right or wrong, how to pick the sample, and what table to record the results in.
Below is part of our company’s [work rules]. Create ten questions employees would realistically ask, with answers based on the text of the rules. Have it say so when something is not covered in the rules. [Paste the rules here]
Suppose document processing costs [$0.01 per page] and we process [3,000 pages] a month. With advanced processing it becomes [$0.03 per page]. Compare the monthly and annual cost of both options in a table, and note that VAT is extra.
Draft the table of contents for an adoption review report comparing Korean and overseas AI vendors. Items: [data storage location], [contracting entity], [pricing structure], [integration with existing systems], [staff in charge]. For each item, also list the questions we must confirm.
Using it with other tools
Linking two or three tools changes your work more than any single tool does.
Read documents → write report
Policy handbook → internal Q&A
Before you start
- Pay-as-you-go pricing means costs can quietly grow during testing. Set limits and alerts first.
- Listed prices exclude VAT. Add tax when you set your approval budget.
- Do not put real customer information in test documents. Measure first with samples where names and amounts have been changed.
- Accuracy varies with document format. Check it with your own company’s documents, not someone else’s success stories.
Pricing: personal use and work use
Plans that individuals pay for are different from plans that organizations contract for by member.
For individual use
Cached input is US$0.06 per 1 million tokens. Solar Pro 3 and Solar Pro 2 are US$0.15 for input and US$0.60 for output
Text-only extraction costs $0.0015 per page, and field extraction starts at $0.04 per page
Per official guidance, 10 free runs per agent. You can test with real documents without registering a card
Using it at work
10% bonus credit on the prepaid amount. Standard rate limits, community support
15% bonus credit on prepaid amounts. One tier higher rate limit, email support
20% bonus credit on the prepaid amount. Highest rate limits, priority support
Custom pricing, with flexible request limits and dedicated support arranged separately, according to the page
Checked 21 Sep 2026 · Amounts shown on Upstage’s official pricing page (in Korean). The page states that 10% VAT is charged separately. Prices change without notice, so check the official page before you pay. Upstage official pricing (Korean) ↗ · Upstage official home ↗
Where Upstage Solar falls short
We don’t just list the positives. Here is what fell short when we used it.
- Prices are listed in US dollars
Although it is a Korean company, prices are shown in US dollars, so for an approval request you have to apply an exchange rate to convert to won and add VAT separately. For whoever builds the budget, that extra step is a hassle.
- You can’t get started on your own
You can sign up on your own, but putting the results to use takes someone who can connect them to internal systems. Without that person, evaluations tend to stall at impressive-looking demos.
- Model names change often
Model generations move up fast, so the model names on the price list change often. A comparison table you prepared last month no longer matches this month’s meeting, so you end up rechecking every time.
Tools to compare
ClaudeThe most talked-about AI→
GeminiGoogle’s all-purpose AI→
inline AIBuilt for official HWP documents→
WrtnKorean AI you can use without logging in→Claude in ExcelAI built into Excel→
ZapierWorld’s largest workflow automation tool→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.
Upstage Solar, built on your organization’s strengths
Send us a message and we’ll reply within 24 hours. We’ll propose a curriculum tailored to your industry and job roles.
Courses in this fieldAI Agents & Work Automation: Hands-On Training