The industrial AI engine Phuc Xin for your company
Dedicated compute and our own generative model. You pay only for minutes of finished output. Onboarding takes 1–3 business days; we work with legal entities under contract only.
One engine for content, documents and automation
Phuc Xin is a large model and runtime built for enterprise workloads. You rent a dedicated instance: compute is never shared with other clients and your data is never used for training.
Text generation
Customer replies, product descriptions, reports, emails and summaries in your brand voice.
Document analysis
Contracts, invoices, specifications and scans: data extraction, clause checks, version comparison.
Multilingual
Chinese, English, Russian and 30+ more languages. Translation that keeps your industry terminology.
AI agents and integrations
Connects to CRM, ERP, knowledge bases and internal APIs. Agents carry out multi-step tasks.
Private deployment
A dedicated cluster in our data center, in your cloud or on your own servers (on-premise).
Domain fine-tuning
Tune the model on your data: terminology, tone of voice, internal policies.
How billing works
The billing unit is one second of finished generated content. Drafts and regenerations you don't accept are free. Usage is metered per second and the report comes with your accounting documents.
- REST API, gRPC and SDK
- Load and spend dashboard
- Request logs kept up to 90 days
- Access keys and staff roles
| Architecture | Mixture-of-Experts, 236B parameters |
|---|---|
| API response time | under 80 ms (median) |
| Interfaces | REST API, gRPC, WebSocket, SDKs for 5 languages |
| Concurrent requests | 20 to 500 req/s depending on plan |
| Languages | Chinese, English, Russian and 30+ more |
| Input formats | text, PDF, DOCX, XLSX, images |
| Deployment | dedicated cloud / private cloud / on-premise |
| Availability | 99.9–99.95% SLA |
| Metering | per second, with monthly report |
API integration in one day
The engine runs as a service inside your infrastructure: your systems call it over a secure API, and we take care of compute, updates and stability.
- REST API and gRPCstandard protocols, OpenAPI docs
- Streaming responsesWebSocket and Server-Sent Events
- SDKPython, Java, Go, Node.js, C#
- Ready-made connectorsCRM, ERP, corporate messengers, knowledge bases
- Access controlAPI keys, roles, IP allowlists, audit logs
# Example request to a dedicated instance
curl https://api.phucxin.org/v3/engine/run \
-H "Authorization: Bearer $PL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"instance": "pl-business-01",
"task": "document.analyze",
"input": { "file_id": "contract_2026_118.pdf" },
"language": "en"
}'
# Response
{ "status": "completed", "latency_ms": 74,
"result": { "risks": 3, "summary": "..." } }Where Phuc Xin delivers results
Proven scenarios that we adapt to each company's processes.
Finance and insurance
Application review, reporting analysis, client replies, compliance checks.
E-commerce
Multilingual product listings, customer support, review analysis.
Manufacturing
Technical documentation, engineering knowledge base, quality control.
Logistics and trade
Invoices and customs declarations, partner correspondence, shipment tracking.
Legal services
Contract review, risk detection, document drafting.
Education and HR
Corporate training, assignment review, candidate screening.
Generation pricing
Prices are per second of generated content, metered to the second. The higher your monthly volume, the lower the price. The package and final price are fixed in the contract.
- 1 dedicated instance
- up to 20 requests per second
- REST API and SDK access
- SLA 99.5%
- email support in business hours
- up to 3 dedicated instances
- up to 100 requests per second
- API, SDK and ready-made connectors
- quarterly fine-tuning
- 99.9% SLA, dedicated account manager
- private cloud or on-premise
- from 500 requests per second
- unlimited fine-tuning
- 99.95% SLA, 24/7 support
- dedicated engineering team
| Condition | Discount |
|---|---|
| from 60 minutes per month | −15% |
| from 120 minutes per month | −20% |
| from 300 minutes per month | −30% |
| 12-month prepaid contract | extra −5% |
Free trial access for companies after a short consultation. Test the engine on your own tasks before signing.
Get trial accessPrices in USD, excluding VAT. Payment by invoice under the contract, with monthly accounting documents.
From request to launch in 1–3 days
Request
Leave your contacts and task description in the form.
Consultation
A manager clarifies volume, scenarios and data requirements.
Trial access
3 free minutes of generation to test the engine on your tasks.
Contract
We fix the plan, minute package, SLA and NDA.
Launch
We issue access keys and help with integration and launch.
Your data stays yours
We never use client data to train shared models, and we sign an NDA before work begins.
Isolation
Each client runs on a dedicated instance with no shared resources.
Encryption
TLS 1.3 in transit, AES-256 for stored logs.
Access control
Roles, IP-restricted keys, audit of every action.
Data retention
Log retention is configurable, down to fully disabled.
Phuc Xin AI
We build and operate our own AI engine and provide it to companies under contract. Our team brings together machine learning engineers, infrastructure specialists and industry consultants.
- Founded
- 2023
- Focus
- Development and rental of an enterprise AI engine
- Working model
- B2B only, by contract, payment by invoice
- Headquarters
- 21F, Tower A, 88 Qianjiang Rd, Binjiang, Hangzhou
- Registration
- Phuc Xin AI Technology Co., Ltd. · Reg. No. 91330108MA2H7K2X8Q
Licenses and certificates
Frequently asked questions
Do you work with individuals?
No. We work only with companies and sole proprietors under contract.
How is output billed?
You pay only for the final duration of the content you accept, to the second. For example, a 2 min 30 s (150 s) video on the Start plan costs $141.
How do we connect the engine to our systems?
Via REST API or gRPC. We provide documentation and SDKs and support your engineers during integration.
Can we pay online?
Payment is made by invoice under the contract. Online card payments are not available.
Can the engine run on our own servers?
Yes, on the Enterprise plan. Hardware requirements are agreed during consultation.
How long does onboarding take?
Trial access is issued within one business day; full onboarding after signing takes 1–3 days.
Request a quote
A manager will contact you within one business day, prepare a commercial offer and open trial access.
- Sales email
- sales@phucxin.org
- Phone / WeChat
- +86 571 8617 2490 · WeChat: phucxin_ai
- Business hours
- Mon–Fri, 9:00–18:00 (UTC+8)