Models and Limits
Public model names, context windows, output limits, concurrency, and unsupported capabilities.
Sovereign exposes a stable public model name:
The underlying reasoning model is managed by Sovereign and may change over time. This allows the platform to improve quality, reliability, latency, and cost without requiring client changes.
Current Capabilities
Sovereign is optimized for:
- technical reasoning;
- cybersecurity analysis;
- code review;
- CVE and exploit triage;
- infrastructure review;
- report writing;
- long-context workflows.
Plan Limits
Limits depend on your plan.
| Plan | Context | Output | Active Requests |
|---|---|---|---|
| Go | 256,000 tokens | 128,000 tokens | 1 |
| Plus | 512,000 tokens | 256,000 tokens | 2 |
| Pro | 1,024,000 tokens | 384,000 tokens | 5 |
Context includes the conversation and attached text context. Output is the maximum generated response size for a single request.
Thinking Mode
Sovereign enables model-side reasoning by default to improve instruction following and complex technical answers.
The private reasoning stream is not exposed to API clients or OpenWebUI users. Only the final answer is returned.
Vision and Images
Sovereign currently does not expose a vision-capable model. Image inputs are rejected explicitly instead of being silently ignored.
Model Details
Sovereign intentionally abstracts the upstream model identity. Clients should depend on the public model contract rather than a provider-specific model name.
Use:
Do not depend on internal routing details.