Get started
Models
What is available, and how to discover it programmatically.
Listing models
GET https://api.civix.com.vn/v1/models
for model in client.models.list():
print(model.id)
{
"object": "list",
"data": [
{ "id": "qwen3.8-27b", "object": "model", "owned_by": "civix" }
]
}
The id values returned here are the only valid values for the model field
of a chat completion request. Treat this endpoint as the source of truth
rather than hardcoding a list — availability changes.
Available now
| Model | id |
Context window | Input | Output |
|---|---|---|---|---|
| CiviX Qwen3.8 27B | qwen3.8-27b |
262,144 tokens | Text, images | Text |
The context window is read from the running server configuration, not estimated.
Choosing a model
With a single model available, the practical choices are about how you call it rather than which one you pick:
- Long documents or large codebases — the 262K context window means you
can often pass material directly instead of building retrieval. Use
"stream": true; see Streaming. - Images — this model accepts image input. See Chat Completions.
- Short interactive replies — non-streaming is fine and simpler.
What we do not publish
We do not publish benchmark scores or throughput figures. Numbers measured on our hardware would not predict what you see under your own workload and concurrency, and quoting them would be misleading rather than useful.