Connect Jan to ChinaGenAI.

Jan Desktop can use a custom OpenAI-compatible endpoint. This guide maps Jan's provider fields to ChinaGenAI and explains what its model-list check does—and does not—confirm.

Add ChinaGenAI as a custom provider

1. Check current service status and site-key availability. Jan requires a key for a remote custom provider.
2. In Jan, open Settings → Model Providers → Add Provider.
3. Choose OpenAI-compatible, then enter the provider details below.
4. Click Create. Jan tries to load the model list from the API base URL.

Enter the API root ending in /v1. Do not append /chat/completions; Jan builds the request path. Use the key issued by ChinaGenAI, not a key from OpenAI or another provider.

Understand Jan's model-list check

When you save a custom provider, Jan requests GET https://chinagenai.com/v1/models. ChinaGenAI lists free while its API pool is available. Seeing the model confirms discovery only; it does not send a chat request or verify that a key can complete one.

If Jan shows no model, first check whether the site pool is available. You can add free manually in Jan's Models section, but manual entry cannot restore a paused service or make an unavailable key work.

What Jan’s “Test keys” result proves

Jan documents its key test as a request to https://chinagenai.com/v1/models. ChinaGenAI's model-list route does not check the Authorization header, so that result can confirm endpoint reachability and model discovery without confirming that the site key is valid. The key is checked when a chat request is sent to https://chinagenai.com/v1/chat/completions. This distinction follows from Jan's documented test and the current public route implementation; no Jan key test or model call was run for this guide.

Check the limits before using Jan

• Short text chats: requests allow up to 6 messages and 6,000 combined input characters. A response is capped at 256 tokens, so long answers can stop early.
• Shared access: each network or verified account allows 20 attempts per UTC day; a key allows one active request. The site allows two active requests and 200 attempts per UTC day across the whole site. Keys expire after 48 hours. Check the live status page for current availability.
• Streaming: the endpoint returns compatible SSE after the upstream response completes; it does not deliver tokens as they are generated.
• Capabilities: this route is text Chat Completions. It does not expose Responses, images or audio. Function-tool fields can pass through the API, but Jan tool use with this pool has not been tested. Do not mark tools or other capabilities as supported unless the selected upstream route is verified.

If Jan cannot connect

• 404: keep the base URL at https://chinagenai.com/v1; remove any extra /chat/completions path.
• 401: use a current ChinaGenAI site key. A site key expires after 48 hours; check access in the console.
• 429: an account, network or shared concurrency limit may have been reached. Wait for an active slot or the next UTC day; replacing a key does not reset the allowance.
• The model is missing: check the API pool first. Jan's /models request and a successful chat are separate checks.

For general client settings, read the OpenAI-compatible connection guide. If you already have a response to inspect, use the local response checker without making another model call.

Official reference

Jan Desktop: Custom Endpoints

Jan's custom-provider instructions were checked against its official documentation on 21 October 2026. This is a written configuration guide; no Jan client session or model inference was run.

Free API access and limits

Check the site free API access and limits before trying to obtain a site key. Read the current status and request limits before using a site key. It is not a source of official provider keys.

ChinaGenAI is not affiliated with the providers named here. This release contains setup guidance and explicitly labeled fictional examples, not claimed live model benchmarks.