Models
zen-guard-gen
8B dense model for safe text generation with built-in guardrails.
zen-guard-gen
Safe Generation
An 8B dense transformer with built-in safety guardrails for content generation. Produces helpful responses while automatically filtering harmful, biased, or inappropriate content at generation time.
Specifications
| Property | Value |
|---|---|
| Model ID | zen-guard-gen |
| Parameters | 8B |
| Architecture | Dense |
| Context Window | 32K tokens |
| Status | Available |
| HuggingFace | -- |
Capabilities
- Safe text generation with built-in content filtering
- Automatic toxicity and bias mitigation
- Harmful content refusal with helpful alternatives
- PII detection and redaction
- Policy-compliant output generation
- Customizable safety thresholds
Usage
API
from hanzoai import Hanzo
client = Hanzo(api_key="hk-your-api-key")
response = client.chat.completions.create(
model="zen-guard-gen",
messages=[{"role": "user", "content": "Write a customer support response for a billing issue."}],
)
print(response.choices[0].message.content)See Also
- zen3-guard -- 4B safety classifier
- zen-guard-stream -- Streaming moderation
- zen4-mini -- 8B general-purpose model