### What happened? After upgrading LiteLLM proxy, the Spend Logs UI frequently shows:

> Request/Response Data Not Available

with instructions to enable:

```yaml general_settings: store_model_in_db: true store_prompts_in_spend_logs: true ```

But these settings are already enabled, and prompt data is still being stored (just not in the expected field).

### Environment - LiteLLM image: `docker.litellm.ai/berriai/litellm:main-latest` - LiteLLM package version in container: `1.82.1` - Postgres: `15.17-alpine` - Redis: `7.4-alpine`

### Relevant config ```yaml general_settings: master_key: os.environ/LITELLM_MASTER_KEY api_key: os.environ/LITELLM_API_KEY database_url: os.environ/DATABASE_URL store_model_in_db: true store_prompts_in_spend_logs: true ```

### Reproduction 1. Start proxy with the above config. 2. Send a normal request via `/v1/chat/completions`. 3. Open Spend Logs UI and click request details. 4. Banner appears for many rows saying request/response data is unavailable.

### What I observed in DB For a fresh request (`request_id = 0852eecf-87a5-40a9-a567-dd552a69ce90`): - `messages` is `{}` - `proxy_server_request` is populated (contains input messages) - …