1 language
1 article delivered
CNY 50.00
Making a brand readable to AI engines has a cost you can calculate in advance. Every capability is priced per use and quoted in Chinese yuan (CNY) — we publish no converted figures, because exchange rates drift. You top up first and use the balance afterwards, and unused balance can be refunded to the original payment channel. If a generation task ultimately fails, the amount already deducted is returned to your balance automatically, with no separate request. All prices are quoted in 1 currency (CNY), with no converted figures published.
From CNY 0.50 for one image to CNY 50.00 for a citation-ready GEO article — every price is published, charged per use, and returned automatically when a task fails. If you want to know whether AI engines can find your content, start with a cost you can predict.
Five billable categories, five units of measurement. All figures are quoted in CNY and charged from the balance in the Shangxun Media GEO Project workspace. 5 categories and 5 units of measurement cover the whole list.
| Capability | Billing unit | Published price | Notes |
|---|---|---|---|
| Image Multi-image generation / AI model / AI outfit change | per generation | CNY 0.50 | Identical for every image tool; one run that outputs several images (multi-image composition) still counts as a single generation 1 generation counts once regardless of how many images it returns. |
| Video Talking-head video / text-to-video / cross-border product explainer / multi-image video | per 5 seconds rounded up | CNY 1.00 | Calculated as ceil(seconds / 5) × CNY 1.00 |
| GEO article | per article | CNY 50.00 | Flat price with no negotiation; each selected language is billed as its own article |
| Task collaboration | per task budget | set by the requesting client | Base reward plus tiered performance bonus; the budget is reserved in advance |
| 3D engineering principles Internal testing, not yet open | per second | CNY 3.00 | Separate rate; not interchangeable with any video unit |
Prices follow the latest display inside the WeChat mini program.
Video is charged per 5-second billing unit, and any part of a unit is counted as a full unit. The table shows the durations used most often.
| Video length | Billing units | Cost |
|---|---|---|
| 5 seconds | 1 unit | CNY 1.00 |
| 8 seconds | 2 units (charged as 10 seconds) | CNY 2.00 |
| 10 seconds | 2 units | CNY 2.00 |
| 15 seconds | 3 units | CNY 3.00 |
| 20 seconds | 4 units | CNY 4.00 |
| 30 seconds | 6 units | CNY 6.00 |
| 60 seconds | 12 units | CNY 12.00 |
Selectable aspect ratios: 1:1 / 3:4 / 9:16 / 16:9. Selectable durations: 5 / 10 / 20 / 30 / 60 seconds (default 10 seconds). Subtitles and voice-over are available in 13 languages — Chinese, English, Japanese, Korean, French, Spanish, German, Thai, Russian, Indonesian, Arabic, Portuguese and Italian. Choosing several languages for one video does not change the price. Target GEO engines — Doubao, Yuanbao, Kimi, DeepSeek, ERNIE Bot, ChatGPT, Perplexity and Gemini — can be ticked at no extra charge.
A GEO article costs a flat CNY 50.00, and each selected language is billed separately — because each language is written as its own article and polished by a native-level model rather than machine-translated.
1 article delivered
3 articles delivered
14 articles delivered (full language coverage)
The 14 supported languages are Chinese, English, Japanese, Korean, German, French, Spanish, Portuguese, Russian, Arabic, Italian, Thai, Vietnamese and Indonesian. Articles come in four length tiers (800 / 1,500 / 3,000 / 5,000+ characters) and six styles (professional review, industry analysis, recommendation, press release, tutorial, comparison). For reference, the GEO Domestic plan costs CNY 3,680 per quarter and includes 4 English articles, distribution to 8 platforms, monthly refresh and citation tracking; the overseas GEO Standard plan costs CNY 3,980 per quarter and includes 12 Chinese and English articles, 14 platforms, an encyclopedia source-material pack and a full monthly report. See what each plan includes →
Before delivery, each article is run through six locally computed editorial checks, and any shortfall comes with a concrete correction. The checks describe writing quality — they are not a performance forecast. 6 checks run before delivery, each with a concrete correction attached.
| Check | What it looks for | Correction when it falls short |
|---|---|---|
| 1. Named quotations | At least two quoted statements attributed to named sources | Insert at least two named quotations |
| 2. Data density | Numbers, percentages and dates that make claims concrete | Quantify qualitative descriptions |
| 3. Cited sources | Checkable external links behind key claims | Embed an external source for each key assertion |
| 4. Self-contained paragraphs | Each paragraph stands alone without back-references | Remove referring words so any paragraph can be extracted on its own |
| 5. Executive summary | The opening answers the core question directly | Move the conclusion to the front |
| 6. Structured hierarchy | H2/H3 levels, tables and a FAQ section | Keep the heading hierarchy and the FAQ block |
A keyword-stuffing check runs alongside these six: repeating the theme word too often loses points and triggers a prompt to substitute synonyms. The scoring behind these checks is an internal editorial rubric, not a customer-facing percentage promise.
For CNY 1.00 you get a measured report on where your brand actually stands in search and inside AI answers. The report contains four deliverable blocks, and nothing beyond them.
Real search-position probes for 8 keywords. Every keyword is returned with your rank and the list of sites that entered the leading positions.
The 10 domains that appear most frequently across the leading results for those same keywords — the occupancy your brand is competing against.
Keywords that do return results and do attract searches, but where you hold no position — demand that exists while your brand is absent from it. A gap keyword therefore has 2 properties at once: provable demand and zero brand position.
Measured mentions across 4 questions — two drawn from the competitive position and two from the opportunity position — reported as a mention rate. The denominator counts only the questions that were actually tested.
You choose one measurement track: Chinese (Doubao / Yuanbao / Baidu) or English (ChatGPT / Perplexity / Google). Each track runs 3 engines, and only 1 track is used per measurement.
Supplying your website URL makes matching more accurate, because a domain match takes priority over a brand-name match. If no site is supplied, the report states plainly that no website was bound, that matching ran on brand name only, and that rankings may therefore read low. 1 bound URL is enough to put domain matches ahead of brand-name matches.
The report does not write articles, does not publish anything automatically, does not ghost-write encyclopedia entries and does not include a full-site audit. Any item that could not be tested is marked as untested, is excluded from the mention rate, and is never estimated or filled in with a zero. The whole probe runs on plain retrieval rather than a large language model, which is why the result comes back quickly. 4 items are explicitly excluded, and untested items are labelled as such rather than estimated.
We do not guarantee that any AI engine will cite you. We do not write or publish encyclopedia entries on your behalf — we prepare a sourced material pack so you can go through each community's own review. We do not disparage competitors: every comparative statement needs its own attributable source. Instead of an outcome guarantee, we provide 1 sourced material pack you can use yourself.
Public research on AI citation supports the reason this report measures structure as well as rankings: verified and structured data accounted for 54.53% of independent citation sources in a 2026 study of 17.2 million citations. This is third-party data, not our own measurement, and it does not predict results for any client. Checkable sources: Writesonic / Ahrefs reporting (2026), GEO measurement benchmarks (2026). Sample sizes and definitions differ between studies.
You can also enter any amount yourself — no less than CNY 1, as a whole number or with two decimal places — and pay through WeChat Pay.
Custom amounts also accepted
Custom amounts also accepted
CNY 138 credited
CNY 328 credited
The system automatically grants free trial credits. They are issued once only and can be used immediately without topping up first; the exact allowance follows the latest display inside the WeChat mini program. Once the trial credits are used up, a top-up is required to keep generating. Registering also currently includes one image and one 5-second video, as displayed in the mini program. Free credits cannot be withdrawn and cannot be converted into balance — they simply expire when used.
Your account balance can be refunded to the payment channel it came from. Submit the request under My → Refund request and we return the amount once it has been verified. For larger amounts or unusual cases, contacting the human support desk directly is faster. Refunds follow 1 route: submit under My, then Refund request.
If a generation task ultimately fails, the system returns the deducted amount to your balance automatically and you do not need to file a separate request. The refund shows up under My → Balance details. 1 automatic refund path applies, with no separate request needed.
Submit your request under My → Invoice request. It requires a receiving email address and an invoice title (the full registered name of the entity). Invoice requests are processed on business days and sent to the email address you provide. Invoicing requires 2 details: a receiving email address and the full registered entity name.
Operated and invoiced by Chongqing Shangxun Media Co., Ltd. (unified social credit code 91500103686203135J). Invoices, contracts and corporate bank transfers all use that name. Prices follow the latest display inside the WeChat mini program.
Worked examples for the most typical workloads. The order page always shows the live amount for your own selection. The order page always shows 1 live figure: the amount for your own current selection.
| Scenario | Usage | Cost |
|---|---|---|
| E-commerce listing: multiple angles of one product | 1 multi-image generation (3 images out) | CNY 0.50 |
| Virtual try-on: white-background garment plus model shot | 3 AI outfit changes | CNY 1.50 |
| Social short video: one 15-second talking-head clip | 15 seconds (3 units) | CNY 3.00 |
| Product explainer: 30-second cross-border walkthrough | 30 seconds (6 units) | CNY 6.00 |
| Multilingual coverage: GEO articles in Chinese, English and Japanese | 3 languages | CNY 150.00 |
| Full overseas GEO, including distribution and monthly maintenance | GEO Domestic, 1 quarter | CNY 3,680.00 |
Statements on this page about capabilities, prices and process are compiled from publicly available product information; available capabilities and prices follow the latest display inside the WeChat mini program. Nothing here is a performance guarantee — see the Terms of service and Privacy policy.
Pricing is broken down to a single image and a single 5-second clip precisely so you can verify the marginal value to your own business at the lowest possible cost, before committing to a plan.
To start: search WeChat for the mini program 「商讯传媒全球品牌推广与策划」(Shangxun Global Brand Promotion & Planning), then open the Shangxun Media GEO Project workspace.