Dli.li AI Docs

English

简体中文

English

简体中文

Toggle theme

FAQ ​

Quick troubleshooting

If you are not sure whether the issue comes from the endpoint, key, model, or billing rules, first return to [Docs Navigation](/en/markdown-examples), [Site Links and Billing Rules](/en/site-links-and-billing), and [API Tokens](/en/guide/console/api-token) to verify the basic setup.

Common entry points ​

Balance and billing ​

What is balance usage and how is it calculated ​

Usage is usually calculated from a combination of model multiplier, group multiplier, and token consumption. In simple terms, different models, different groups, and different output lengths can lead to different charges. The whole site settles in CNY, and USD-based official prices are converted using the current exchange rate.

How are model multipliers calculated now ​

The old 0.15 base multiplier for foreign models has been removed. All models now use their normal multipliers and the whole site settles in CNY:

  • Model multipliers are applied directly, without an extra 0.15
  • The whole site uses CNY settlement
  • The RMB / USD ratio follows the site’s current exchange rate

Example:

  • If a model’s official price is 1 USD, the model multiplier is 2, and the exchange rate is 7.2
  • Then the final billed price is 1 × 2 × 7.2 = 14.4 CNY

If pricing still looks confusing, start with Site Links and Billing Rules.

Why do I still see insufficient balance even though my account balance is enough ​

The common cause is not the total account balance, but the token itself having a separate usage cap. Check these items together:

  • Whether the account balance is sufficient
  • Whether the current API token has its own limit configured
  • Whether the model is allowed for the current token

Models and requests ​

Why does it say there are no available models or no available channel ​

For regular users, check these items first:

  1. Whether your current account has access to that model.
  2. Whether the current token limits which models can be called.
  3. Whether the model name entered in the client exactly matches the model name exposed by the site.

If those all look correct, inspect Usage Log for the more specific error.

Where should I look up model names ​

Go to Profile to view the models available to your account, or call GET /v1/models.

Client integration ​

Which domain should I use ​

  • Console pages, the model gallery, the user agreement, and the privacy policy can all be opened through the same path on all three domains
  • Login, payment, subscription, recharge, and external account linking should use the main site https://dli.li
  • For regular API traffic, the official default API is https://api.dli.li/v1
  • Official backup API: https://api.dlizz.com/v1

A third-party client reports Failed to fetch ​

Check these items first:

  1. Whether the endpoint is correct and usually includes /v1.
  2. Whether the API key was copied completely.
  3. Whether the current browser or client environment requires HTTPS.
  4. Whether you accidentally entered the upstream vendor’s official endpoint instead of your own Dli.li AI endpoint.

Why can I still not connect even after filling the address and key ​

The three most common causes are:

  • The address is missing /v1
  • The model name is wrong
  • The API key does not belong to the current account

Try the same model first in Playground or with curl to confirm the basic request path works.

Why does a Claude client fail on /messages ​

Because most models on this site do not support /messages, and only some Claude models do. Prefer the compatible calling path by default.

Why can I not see multiple subscription model groups at the same time ​

That follows the subscription rules:

  • You cannot subscribe to multiple plans at the same time
  • Only the models from the current subscription are shown
  • A new subscription overrides the previous one

Logs and troubleshooting ​

Which page should I use to troubleshoot ​

What should I do if billing looks wrong ​

Confirm these four items first:

  1. Which model you used.
  2. What the current model multiplier and group multiplier are.
  3. Whether the request used streaming, long output, or multi-turn context.
  4. What the site’s current exchange rate is.

Then compare the actual request in Usage Log.

What if the issue is still not solved ​

If the information above still does not solve it, prepare these details before reporting the issue:

  • The client or program you are using
  • The endpoint address
  • The model name
  • The original error message
  • A screenshot of the matching logs and time

That usually makes the problem much easier to locate. If you need human support, visit Contact.