Problem:
The current /v1/models API returns all models (active, deprecated, preview, legacy) with no clear distinction.
Example current response:
{
"id": "gpt-4.5-preview",
"created": 1690000000,
"owned_by": "openai"
}
Issue:
- No way to know if a model is active or deprecated
- Developers must manually filter using naming conventions
- Leads to confusion and risk of using outdated models
Context:
OpenAI docs mention deprecations separately, but the API itself does not expose lifecycle info.
Suggested improvement:
{
"id": "gpt-5.2",
"lifecycle": "active", // active | deprecated | retired | preview
"created": 1690000000,
"owned_by": "openai"
}
Impact:
- Cleaner developer experience
- Safer production usage
- Eliminates fragile string-based filtering
- Aligns with standard API lifecycle practices
Current workaround:
Developers must maintain allow-lists or filter by name patterns (e.g. “preview”, “legacy”), which is brittle.
Ask:
Can we expose lifecycle metadata or a “recommended models” subset directly in the API?