Raw text post from archive
AI Models & Assistants / Knowledge library
text-moderation-stable
Previous generation text-only moderation model
Why learn text-moderation-stable?
Offers the documented OpenAI model capabilities through hosted.
The problem it solves
Provides model inference for supported text inputs.
Where it came from
text-moderation-stable: Previous generation text-only moderation model Published aliases: text-moderation-007. Input: text. Output: text. Checked 2026-10-13. Model capabilities are distinct from tools available in a particular agent.
Build your understanding
Work through these exercises in order. Try the task before opening a reference.
- Explain what text-moderation-stable does.
- Identify a use case for text-moderation-stable.
- Describe a limitation of text-moderation-stable.
- Name a compatible tool for text-moderation-stable.
- Compare text-moderation-stable against an alternative.
Documentation and sources
- Primary source / model card (developers.openai.com)
Catalog research notes
- See the AI Workbench for source dates, pricing evidence and configuration details.
Connect the ideas
Built on
Real World Example
Legacy Forum Content Governance
A community-driven historical archive platform hosting millions of legacy text posts.
The business problem
The platform struggles with high volumes of user-generated content that may violate community safety guidelines, requiring a consistent, cost-effective way to flag potentially harmful text without migrating to newer, more complex model architectures.
How this topic solves it
By integrating the text-moderation-stable model, the platform can programmatically scan incoming and existing text submissions against safety categories. This model provides a stable, locked-in snapshot (text-moderation-007) that ensures consistent classification behavior for legacy datasets, avoiding the variability that might arise from frequent model updates.
- Identify the specific text fields in the legacy database requiring moderation.
- Configure the application to route text payloads to the v1/moderations endpoint.
- Utilize the text-moderation-007 snapshot to ensure consistent classification logic.
- Implement a conditional logic gate to flag or block content based on the model's returned categories.
- Log the moderation results for audit trails and manual review queues.
This diagram illustrates the flow of text content from the user archive through the moderation model and into the final decision state.
Check if content is text-only
text-moderation-007
Flag or Approve
- User Input → submit → Validation
- Validation → valid text → Moderation API
- Moderation API → classification results → Decision Engine
Expected business outcome
The platform achieves a predictable and automated safety filtering process that maintains historical consistency across its archive.
Limitations and checks
The primary limitation is that text-moderation-stable is a text-only model and cannot process multimodal content like images or audio. To check for this, developers must validate input types before calling the moderation endpoint; if the input contains non-text media, the system should trigger a fallback error handler.