Free OpenCode Zen models consistently return HTTP 429 despite being listed as available.

Description

Description

I am using OpenCode with the OpenCode Zen free models for a real
structured-extraction workload.

The free models are shown as available in the OpenCode model list, but
the API consistently returns HTTP 429 (rate limit) before generating any
response.

This is preventing us from using the free Zen models for an external
application integration.

Models tested

• deepseek-v4-flash-free
• mimo-v2.5-free (if applicable; otherwise remove this line)
• Other free Zen models shown in the current model selector

The most reproducible case so far is:

deepseek-v4-flash-free

Steps to reproduce

1. Obtain an OpenCode Zen API key.

2. Call the OpenAI-compatible endpoint:

   https://opencode.ai/zen/v1/chat/completions

3. Use the model:

   deepseek-v4-flash-free

4. Send a normal chat-completions request.

Example:

from openai import OpenAI

client = OpenAI(
    base_url="https://opencode.ai/zen/v1",
    api_key="YOUR_ZEN_API_KEY",
)

response = client.chat.completions.create(
    model="deepseek-v4-flash-free",
    messages=[
        {
            "role": "user",
            "content": "Return a short JSON object with your model name."
        }
    ],
    temperature=0.2,
)

print(response)

### Plugins

_No response_

### OpenCode version

_No response_

### Steps to reproduce

_No response_

### Screenshot and/or share link

_No response_

### Operating System

_No response_

### Terminal

_No response_