Use Zenifra AI in OpenCode
Zenifra AI is OpenAI-compatible and works in OpenCode through the official plugin zenifra/opencode-zenifra. The plugin brings the catalog models into the OpenCode model picker and keeps the list up to date on its own.
Get the key
The key comes from the Zenifra console:
- AI plan per member: in AI Models > Plans > My plan, in the Connect your tools section. The key starts with
zai_. See AI plans per member. - On-demand AI: in AI Models > On demand > API Keys. See Getting started with AI.
Treat the key like a password: every use made with it counts toward the plan or the organization's credit.
Install the plugin
In the terminal:
opencode plugin add github:zenifra/opencode-zenifraTo check the installation:
opencode plugin listConnect the key
In OpenCode, use the Connect an integration action (shortcut ctrl+a in the model picker) or the /connect command, choose Zenifra and paste the key. OpenCode stores the key, so it does not need to go in the config file.
If you prefer, use the ZENIFRA_AI_API_KEY environment variable:
export ZENIFRA_AI_API_KEY="your-key"
opencode --standaloneIf OpenCode runs as a service, set the variable on the service:
opencode service set env ZENIFRA_AI_API_KEY "your-key"If a key is connected through OpenCode, it takes precedence over the environment variable.
Pick the model
Open the picker with /models. Zenifra models show up by name, for example Qwen3.8 Flash, under the Zenifra provider. The choice applies to the current session.
For a run from the terminal, use the catalog ID:
opencode run --model zenifra/qwen3.8-flash "Explain the structure of this project."Usage limits
On an AI plan, the key has the 5-hour window and the weekly one; see how the limits work. On on-demand AI, the key quotas and the organization's credit apply.
When a request does not fit the limit, the response says why and what to do, and OpenCode shows that message. For example:
The 5-hour usage limit of this key is used up. It resets at 2026-11-10 01:00 UTC.Other OpenAI-compatible clients
Any OpenAI-compatible client works with three settings:
- base URL:
https://ai.zenifra.com/v1; - key: the same plan or on-demand key;
- model: the catalog ID, such as
zenifra/qwen3.8-flash.
The available IDs are in the public catalog https://ai.zenifra.com/v1/models.
Next steps
Last updated on
AI plans per member
AI plans with a fixed price per member: Lite, Standard and Premium, the member's own API key, 5-hour and weekly limits and billing prorated to the cycle.
Getting started with AI
Create an AI API key on Zenifra, set its daily and monthly quotas and allowed models, and make your first OpenAI-compatible request with the Python SDK.