OpenAI
Leveraging OpenAI within Lakehousecat provides AI capabilities for Chat and Speech-to-Text (STT) tasks. This page describes how to configure OpenAI as a Provider Model.
Prerequisites
- Access to Lakehousecat as an Administrator
- An active OpenAI API Key
Configuration
Navigate to Workspace → Models → Provider Models and click the + icon.
Step 1: Select Provider Type
From the Provider Type dropdown, select OpenAI.
Step 2: Select Model Type
Choose the appropriate model type:
| Model Type | Description |
|---|---|
| Chat | Conversational AI (e.g. gpt-4o, gpt-4o-mini) |
| STT | Speech-to-Text (whisper-1, gpt-4o-transcribe, gpt-4o-mini-transcribe) — uses the same API key as Chat, no extra setup |
Step 3: Enter Configuration Details
| Field | Description |
|---|---|
| Configuration Name | A unique name for this configuration (e.g. openai-gpt4o-chat). |
| API Key | Your OpenAI API key. Required for authentication. |
| Model ID | Select the specific model from the dropdown (e.g. gpt-4o, gpt-4o-mini). |
Model ID accepts any OpenAI model identifier as free text — see Tested Configurations below for models explicitly verified by the Lakehousecat team.
| Field | Description |
|---|---|
| Base URL (optional) | Override the default OpenAI API base URL if using a custom endpoint. |
Step 4: Save
Click Save. After saving, you can optionally set this configuration as the default UI model or default Backend model.
Tested Configurations
Stand: 2026-07-20. This reflects what Lakehousecat has actually run and observed, not a benchmark or performance ranking — see There is no single "right" provider for why we don't publish comparative scores.
| Model ID | Chat | Chart Generation | Notes |
|---|---|---|---|
gpt-5.6-luna | ✅ | ✅ (recommended starting point, ~32s/chart, lowest cost of the tested set) | Requires the tool-binding fix Lakehousecat ships (reasoning_effort forced off for tool calls) |
gpt-5.6-sol | ✅ | ✅ (~39s/chart) | — |
gpt-5.6-terra | ✅ | ✅ (~39s/chart, lowest chat latency of the tested GPT-5.6 variants) | — |
gpt-5.5 | ✅ | ✅ (~36s/chart) | — |
o3 | ✅ | ✅ (~45s/chart) | Reasoning model, slightly slower than the GPT-5.6/5.5 chat models |
o4-mini | ✅ | ⚠️ usable with caution (~110s/chart) | Noticeably slower than the standard chat models; only worthwhile if you specifically need its reasoning |
o3-mini | ✅ | ❌ not recommended for chart generation | Latency made it impractical for interactive chart generation in testing |
gpt-5.5-pro | ✅ (Chat only) | ❌ Not available | Enforced by the backend — "pro" reasoning models are blocked for chart generation regardless of configuration. See Cost Awareness & Chart-Generation Restrictions |
o4-mini-deep-research | ❌ | ❌ Not supported | Only reachable via OpenAI's /v1/responses API, which Lakehousecat does not implement — this is a technical incompatibility, not a quality judgment |
Model Selection & Cost Considerations
For chart generation, choose a standard GPT-5.x chat model (e.g. gpt-5.6-luna) — fast, reliable, and considerably cheaper than the -pro and deep-research variants for this structured, tool-driven task.
-pro reasoning models (e.g. gpt-5.5-pro) and deep-research models are not available for chart generation. This is enforced by the backend, not just a suggestion — attempting to use one as a Custom Model's base model, or to trigger chart generation with one, is rejected with a message pointing you to a lighter alternative. The same models remain fully usable as standalone Chat provider models; only the chart-generation use case is restricted. See Cost Awareness & Chart-Generation Restrictions for the full picture.
Best Practices
- Use a descriptive configuration name that includes the model name and type.
- Separate configurations for Chat and STT allow independent management of each capability.
- For chart generation, prefer a standard GPT-5.x chat model — not the
-proor deep-research variants, which are far more expensive and slower without improving the result for this structured task, and which are blocked for chart generation regardless. See Choosing a Model for Chart Generation.
Troubleshooting
- Validation Errors: Verify the API Key for accuracy.
- Model Access Issues: Ensure your OpenAI account has access to the selected model.