Use Cline's OpenAI Compatible provider to connect an OpenAI-format Chat Completions service. For ChinaGenAI, the settings are a custom API base, your site key and the free route label. This gives you the connection values; it does not certify every Cline agent feature or upstream model.
The site pool has daily and per-key concurrency limits. Check current API status and site key access first. The limits below are small enough that this setup is for a short, non-destructive trial, not a long coding session.
Use these three Cline settings
| Cline field | Value |
|---|---|
| API Provider | OpenAI Compatible |
| Base URL | https://chinagenai.com/v1 |
| API Key | Your ChinaGenAI site key, entered in Cline locally |
| Model ID | free |
Enter the API root ending in /v1, not the full /chat/completions route. The request path for this service is POST https://chinagenai.com/v1/chat/completions. Use Cline's custom compatible provider, not its first-party OpenAI provider.
Set it up in Cline
- Open Cline settings and choose OpenAI Compatible under API Provider.
- Enter the Base URL above and the site key from your ChinaGenAI account.
- Enter
freeas the model ID. If a model list appears, the site exposes this label while its pool is enabled; a listed ID only confirms discovery, not a successful chat. - Keep the first task short and non-destructive. Do not put the site key in a checked-in project file, prompt, screenshot or shared log.
Keep image and computer-use capabilities off for this text-only route. The gateway accepts OpenAI-style function-tool fields, but the selected upstream model's tool behavior has not been verified with a live Cline run. Cline's official settings also expose output-token and context-window fields; do not configure them as if they override the site's limits.
What the current limits mean
- Input: the serialized request, including messages, parameters and tool schemas, is capped at 6,000 characters. Cline-generated context may exceed that before you add your task.
- Output: the gateway caps a response at 256 tokens. That is too short for many code edits or multi-step agent tasks.
- Access: one key is limited to 20 requests per day and one concurrent request; the whole site is limited to 200 requests per day and two concurrent requests. Site keys expire after 48 hours.
- Features: text requests and the Chat Completions route are supported. The gateway can pass function-tool fields, but model-specific tool use is not guaranteed.
These are the gateway limits checked on 21 October 2026. Recheck site status because availability and limits may change. If Cline reports an oversized request, reduce the task or use another provider with limits suited to coding-agent context.
If Cline does not connect
- 404: confirm the Base URL ends in
/v1and does not already include/chat/completions. - 401: confirm you used a ChinaGenAI site key, not an OpenAI key. Check whether key access is available in API access.
- Model is missing: type
freeexactly and check pool status. - Tool calls, long prompts or code output fail: check input/output limits and the selected upstream model. An OpenAI-compatible setting does not mean every optional feature works.
For a standard Chat Completions client, see the API connection settings. For other coding assistants, compare the Continue and OpenCode setup guides.
References
Cline: OpenAI Compatible provider settings
Cline community discussion: Chat Completions endpoint selection
Cline configuration guidance checked 21 October 2026. ChinaGenAI settings and limits were read from the public site. No Cline session or model inference was run for this guide.
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.