FAQ#
Answers to the most common questions from users
š Getting Started#
How do I get started with Nexconn AI services?#
Getting started is simple, just 4 steps:Step 1: Register an accountClick the Register button and fill in basic information.
Verify your email or phone number.
Individual users: Provide ID card information
Enterprise users: Provide business license and other documents
Verification is required before purchasing services
Log in to the Nexconn console, select Large Model Services > Service Purchase.
After purchasing and topping up, enter the large model console.
Refer to the documentation sample code.
Or visit the model marketplace to browse available models.
I don't know how to code, can I still use Nexconn AI?#
Absolutely! Nexconn AI offers multiple ways to use:ā
No coding required#
Third-party clients: Use tools like ChatBox, Cherry Studio, etc., configure Nexconn API and start using
API testing tools: Use Postman and other tools to test the API
š» Better if you know coding#
API integration: Integrate AI into your own applications
Custom development: Build your own AI assistant
Batch processing: Automate processing of large volumes of tasks
Which AI models are supported?#
Nexconn provides a rich selection of models, supporting 70+ top global large models:| Model Series | Representative Models | Features |
|---|
| GPT Series | GPT-5.2 | OpenAI's latest model |
| Kimi Series | Kimi | Ultra-long context, Chinese optimized |
| DeepSeek Series | DeepSeek-V3.1 | Cost-effective, strong coding ability |
| Tongyi Qianwen | Qwen | Multimodal, good Chinese understanding |
| Doubao Series | Doubao | Fast response, low cost |
| Zhipu AI | GLM 4.7 | Multimodal, tool calling |
| Minimax | Minimax M2 | Strong creative generation and conversation |
| Open Source Models | GPT-OSS-120b/20b | Open source, controllable, high flexibility |
Model library is continuously updated, more models coming soon! Visit Model Marketplace to see the complete list.
š° Billing & Pricing#
How is billing calculated? What are the pricing standards?#
Nexconn provides pay-as-you-go billing:Pay-as-you-go#
Charged based on actual token usage
Cost = Input tokens Ć Input unit price + Output tokens Ć Output unit price
Users with unstable usage
Users seeking flexibility
Enterprise users can contact sales for bulk discounts
How are tokens calculated? How many characters is 10,000 tokens approximately?#
Token calculation rules#
| Language Type | Calculation Method | Example |
|---|
| Chinese | Usually 1-2 Chinese characters = 1 token | "The weather is nice today" ā 6 tokens |
| English | Usually 1 word = 1-2 tokens | "Hello World" ā 2 tokens |
10,000 tokens is approximately equivalent to#
| Content Type | Approximate character count/content volume |
|---|
| Chinese text | 7,000 - 10,000 characters |
| English text | 5,000 - 7,500 words |
| Novel | About 15-20 pages (A4 paper) |
| Code | About 400-600 lines of code |
Example: A conversation containing a 100-character question and a 500-character answer consumes approximately 600-800 tokens.
Are prices the same for different models?#
No! Pricing varies for different models:Reasons for price differences:š° Model cost: Models with larger parameter counts have higher computational costs
šÆ Capability differences: More capable models typically have higher prices
š Different sources: International and domestic models have different pricing strategies
š Market positioning: Premium models vs. economy models
Most expensive: Current top domestic and international models
Medium: Mainstream models like Kimi, Tongyi Qianwen, etc.
Economy: Cost-effective models like DeepSeek, Doubao, etc.
Cheapest: Open source small models
Recommendation: Choose appropriate models based on task complexitySimple tasks (like classification, summarization) ā Use economy models
Complex tasks (like deep reasoning, code generation) ā Use premium models
š Security & Privacy#
Is my data secure? Will it be used to train models?#
ā
Data security guarantees#
š End-to-end encryptionAll data transmission uses HTTPS encryption to prevent man-in-the-middle attacks
Nexconn does not store your conversation content and sensitive information
All API calls have detailed logs for traceability
Complies with data security and privacy protection laws and regulations
Clear commitment: Data you submit through Nexconn AI services will not be used to train models!
What should I do if my API Key is leaked?#
ā ļø Take the following measures immediately:Step 1: Immediately disable the leaked keyStep 2: Generate a new keyCreate a new API Key and update it in your application.
Check for abnormal calls and assess losses.
Log in to the console to view usage statistics.
ā Do not hardcode API Keys in your code.
ā Do not upload API Keys to public Git repositories.
ā Use environment variables or configuration files to store keys.
ā Rotate keys regularly.
ā Use different keys for different projects.
āļø Technical Issues#
What should I do if API calls are slow?#
Possible causes and solutions#
High latency due to distant servers
Choose nearby service regions
Check local network quality
Slow processing due to overly long context or input text
Streamline input content, remove irrelevant information
Use summarization to compress long text
Process large amounts of data in batches
Long computation time for models with large parameter counts
Use lightweight models for simple tasks
Use streaming output for faster first-token response
Adjust max_tokens parameter to limit output length
Request queuing during peak usage periods
Use during off-peak hours
Use asynchronous calling methods
If none of the above methods work, please contact technical support for diagnosis!
What does OpenAI API compatibility mean?#
"OpenAI API compatibility" means if you've previously used OpenAI services (like ChatGPT API), you can seamlessly switch to Nexconn with almost no code changes!Practical advantages#
ā
Low migration cost: Only need to modify API address and key
ā
Low learning curve: Use the same documentation and examples
ā
Rich ecosystem: Can use OpenAI's third-party tools
Switching example#
šÆ Application Scenarios#
Can it be used for commercial projects?#
ā
Absolutely!#
Nexconn AI services support commercial use, and you can integrate them into various commercial products and services.Common commercial application scenarios#
š¼ Enterprise internal systemsIntelligent customer service systems
Automatic document generation
Intelligent education platforms
Product description generation
Personalized recommendations
Legal document assistants
Medical consultation support
Comply with service agreements and terms of use
Must not be used for illegal purposes
Certain industries (such as healthcare, finance) need to pay attention to compliance requirements
AI-generated content should be reviewed by humans before publication
Who owns the copyright of AI-generated content?#
Copyright ownership#
ā You own the usage rights of generated contentCan freely use, modify, and commercialize AI-generated content
According to laws in different countries/regions, there is no unified conclusion on copyright ownership of AI-generated content
Make appropriate modifications to AI-generated content to enhance originality
1.
Use AI-generated content as creative assistance, not direct use
2.
Conduct human review and polishing of important content
3.
Consult professional legal advice for critical commercial scenarios
4.
Label content as AI-assisted generation (when necessary)
Can it process real-time data?#
ā
Supported real-time scenarios#
Real-time conversation: Supports streaming output, returns as it generates
Web search: Obtain real-time web information through MCP tools
API data: Call external APIs to obtain real-time data (through web search)
Real-time analysis: Analyze and process real-time incoming data
ā ļø Limited scenarios#
Model knowledge cutoff date: Model training data has time limitations (e.g., 2024)
Unconfigured tools: Cannot actively obtain external real-time data without configured MCP tools
Proprietary databases: Cannot directly access your private databases (need to configure through tools)
Use web search functionality to obtain latest information
Connect to real-time data sources through MCP protocol
Provide real-time data in prompts as context
Example: To have AI analyze today's stock prices, first obtain real-time stock price data through an API, then provide it to the AI in the prompt for analysis.
š Troubleshooting#
What should I do if API calls return errors?#
Common errors and solutions#
Cause: API Key is incorrect or expired
ā Check if API Key is correct
ā Confirm key has not been disabled or deleted
ā Regenerate key and update
Cause: Exceeded request rate limit
ā Reduce request frequency
ā Implement request retry mechanism (exponential backoff)
ā Contact customer service to increase quota
Cause: Request parameter format error
ā Check if JSON format is correct
ā Confirm all required parameters are complete
ā Refer to API documentation to verify parameter types
Cause: Server internal error
ā Check service status page
ā If it persists, contact technical support
Cause: Insufficient account balance or resource package
ā Top up account or purchase resource package
ā Check billing details
If you encounter errors you cannot resolve, please save complete error information and request logs, and contact technical support for help!
Why are AI responses inaccurate?#
Possible causes and improvement methods#
Cause 1: Prompt not clear enough| Type | Example |
|---|
| ā Vague prompt | "Write something" |
| ā
Clear prompt | "Please write a 500-word product introduction for a smartwatch targeting young people" |
Cause 2: Lack of context informationAI lacks necessary background knowledge
ā Provide sufficient background information in the prompt
ā Upload relevant documents as reference
ā Use multi-turn conversations to gradually supplement information
The model used is not suitable for the current task
ā Use more powerful models for complex tasks
ā Try different models and compare results
Model doesn't know the latest information
ā Use web search functionality
ā Provide latest data in the prompt
ā Explicitly tell AI that latest information is needed
Tip: If the answer is unsatisfactory, you can ask AI to "think again" or "answer again", sometimes you'll get better results!
Recommended method, fast response time
Real-time response during business hours
Chat window in the lower right corner of the official website
Exclusive channel for enterprise customers
Contact your account manager
When submitting issues, please provide:Detailed problem description
Error message screenshots or logs
Model and parameters used
Account information (do not include keys)
š Other Questions#
Are there usage limits?#
The following limits exist:| Limit Type | Description | How to increase |
|---|
| Request rate (QPS) | Requests per second limit | Contact customer service to increase |
| Context length | Maximum tokens for single input | Use models that support longer context |
| Concurrent connections | Number of simultaneous requests | Contact customer service to increase quota |
| Output length | Maximum tokens for single generation | Configure through parameters (max_tokens) |
| Content restrictions | Prohibited illegal content | Cannot be increased |
Individual users vs. Enterprise users:Individual users have basic quotas
Enterprise users can apply for higher quotas
For large-scale use, recommend contacting sales for dedicated solutions
Is batch processing supported?#
ā
Yes!#
Nexconn AI provides batch processing capabilities, suitable for large-scale data processing scenarios.Batch processing methods:Loop through multiple requests in code
ā Simple to implement ā ļø Watch for rate limits
Process multiple data items in one request
ā More efficient ā Lower cost
Contact sales for custom batch processing solutions
ā Dedicated quota ā Priority processing
Batch document translation
Large-scale data classification and labeling
Massive text summarization generation
Can it be used offline?#
ā Cloud services do not support offline use#
Nexconn AI is a cloud-based inference service that requires internet access.
Summary#
This FAQ covers common questions about Nexconn Large Model API services. If you have other questions:The Nexconn AI team is committed to providing you with quality AI services and technical support!
