### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

**Version:** `docker.litellm.ai/berriai/litellm-database:main-stable` (June 2026)

### What happened?

`POST /model/new` returns HTTP 200 and the model is accessible in memory, but the response contains `db_model: false` and nothing is written to `LiteLLM_ProxyModelTable`. `STORE_MODEL_IN_DB=True` is set as an environment variable and `store_model_in_db: true` is in the config. There is no error in the response and no log warning that the write was skipped.

The model disappears on container restart because it was never persisted.

### Steps to reproduce

1. Start the proxy with `STORE_MODEL_IN_DB=True` and `store_model_in_db: true` in `litellm_config.yaml`. 2. `POST /model/new` with a valid model payload and master key auth. 3. Check the response body: `db_model` is `false`. 4. Run `SELECT COUNT(*) FROM "LiteLLM_ProxyModelTable"` against the connected Postgres. Count is 0. 5. Restart the container. The model is gone.

### Expected behaviour

Either the model gets written to `LiteLLM_ProxyModelTable` and the response reflects that, or an explicit error is returned…