AI Models settings
AI Models configures chat, embedding, image generation, APIM, multi-endpoint routing, and model endpoint identity behavior.
AI Models settings
What this group controls
AI Models configures chat, embedding, image generation, APIM, multi-endpoint routing, and model endpoint identity behavior.
Why it matters
Model endpoints are production dependencies for every generated answer, embedding, and image. Keep deployment names, API versions, and authentication choices aligned with the Azure resources operators support.
Video planned
AI Models settings walkthrough
Recording planned. Walk through each tab in the AI Models group and explain when to change each setting.
To publish this video:
- Upload the recording to YouTube or Microsoft Stream.
- Save a poster frame as
docs/images/video-posters/admin-ai-models.png. - Add the watch URL to
this slotindocs/_data/media.yml.
Before you change anything
- Provision Azure OpenAI, APIM, and image resources before pointing SimpleChat to them.
- Choose key or managed identity authentication and grant required permissions.
- Name fallback deployments deliberately so background tasks use approved models.
Model Endpoints
Model Endpoints
The Model Endpoints section belongs to the Model Endpoints tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.
Verified model capacity
Capacity overrides implemented in version: 0.261.035 (application version in
application/single_app/config.py).
Use Advanced endpoint capacity for verified defaults shared by the models on an endpoint. Use Advanced model capacity in an individual model row when a deployment has different limits or an arbitrary deployment name needs an exact catalog identity. These controls are also available in authorized personal and group workspace endpoint editors.
| Field | Meaning | Scope |
|---|---|---|
| Context Window (tokens) | Shared total that input and generation consume together. | Endpoint and model; contextWindow |
| Input Token Limit (tokens) | Independently documented maximum input, not a synonym for context window. | Endpoint and model; inputTokenLimit |
| Output Token Limit (tokens) | Hard provider generation ceiling, not the amount requested on each call. | Endpoint and model; outputTokenLimit |
| Catalog Model ID | Actual published model ID behind an arbitrary deployment alias; a display name is not authoritative. | Model; catalogModelId |
| Model Version | Exact deployed model version or snapshot, separate from the endpoint’s API version. | Model; modelVersion |
| Token Limit Provider | Hosting profile for the verified limits: Auto / inherit, Azure, OpenAI, Anthropic, Google, Vertex AI, xAI, Publisher, or Custom. This does not reroute requests. | Endpoint and model; tokenLimitProvider |
| Output Token Accounting | Whether the allowance covers total generation including reasoning, visible output only, or an unknown accounting contract. Inherit uses the next applicable source. | Endpoint and model; outputTokenAccounting |
| Response Length | Per-request generation allowance for standard chat, kept separate from model capacity. | Model; responseLength |
Each capacity inherits independently: model override -> endpoint override ->
exact catalog model. Leave a field blank to inherit; clearing an existing
override saves null. A missing or unknown maximum does not mean unlimited
capacity. Only enter positive whole numbers up to 9007199254740991; fractions,
scientific notation, negative values, and zero are rejected.
Server-side validation also rejects invalid overrides before saving the submitted
endpoint configuration or credentials. Personal/group saves return HTTP 400 with
error_code: model_context_invalid and a safe field message. The admin settings
form returns to the settings page with the same message instead of saving the bad value.
Independent input and output maxima need not fit simultaneously inside the context window. The actual request must still fit its shared context and any independent ceilings. Verify deployment/provider/version specifications before overriding limits, especially for custom gateways, on-premises deployments, or a different hosting cloud. Select Total generation only when the provider/API actually includes reasoning and other generated tokens in its allowance.
In Admin Settings, Save Endpoint updates the form; save the main settings form to persist it. Workspace editors persist through their existing scoped save routes. See declaring verified model capacity for the workflow and validation checks.
Chat Model
The Chat Model section belongs to the Model Endpoints tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.
Settings
| Setting | What it does | Default | Notes |
|---|---|---|---|
| Enable multi-endpoint model management | Provides the endpoint or route SimpleChat uses for this service. | Off | enable_multi_model_endpoints; capability toggle |
| Search Agents | Defines behavior for the related admin workflow; verify the affected feature after saving. | N/A (runtime control) | Runtime UI control |
| Filter | Defines behavior for the related admin workflow; verify the affected feature after saving. | N/A (runtime control) | Runtime UI control |
| Enable header | Provides the endpoint or route SimpleChat uses for this service. | Off | model_endpoint_identity_header_enabled |
| Header Name | Provides the endpoint or route SimpleChat uses for this service. | Not specified in defaults | model_endpoint_identity_header_name |
| Identity Value | The selected identity is HMAC-hashed before leaving SimpleChat. Missing identity values omit the header. | Not specified in defaults | model_endpoint_identity_header_value_type |
| Default model for fallbacks and new chats | Used for tasks such as conversation summarization and agent fallback. When new conversation defaults are enabled, this model also replaces the user’s last selected model for new non-agent chats. | Not specified in defaults | Runtime UI control |
| Apply to new chats | Starts new non-agent conversations with the admin default model instead of each user’s last selected chat model. | Off | enable_default_model_for_new_conversations; capability toggle |
| Default reasoning effort | Sets the reasoning effort sent with new non-agent conversations when the admin default is enabled. | Model default | default_reasoning_effort |
| Use APIM instead of direct to Azure OpenAI endpoint | Exposes the capability after required services, permissions, and rollout policy are ready. | Off | enable_gpt_apim; capability toggle |
| Azure OpenAI Endpoint | Provides the endpoint or route SimpleChat uses for this service. | Empty | azure_openai_gpt_endpoint |
| Authentication Type | Chooses whether SimpleChat authenticates to this service with a key, managed identity, or another supported method. | key | azure_openai_gpt_authentication_type |
| Subscription ID | Defines behavior for the related admin workflow; verify the affected feature after saving. | Empty | azure_openai_gpt_subscription_id |
| Resource Group | Defines behavior for the related admin workflow; verify the affected feature after saving. | Empty | azure_openai_gpt_resource_group |
| Azure OpenAI GPT Key | Provides the secret credential used when the selected authentication mode requires one. | Empty | azure_openai_gpt_key |
| Azure OpenAI API Version | Pins the service API version SimpleChat sends with requests for this feature. | 2024-05-01-preview | azure_openai_gpt_api_version |
| Azure APIM Endpoint | Provides the endpoint or route SimpleChat uses for this service. | Empty | azure_apim_gpt_endpoint |
| Azure APIM API Version | Pins the service API version SimpleChat sends with requests for this feature. | Empty | azure_apim_gpt_api_version |
| Azure APIM Deployment | Each model defined here will be available in the Chat UI as an option for the User. You can include multiple models seperated by a comma (example: gpt-4o, o-1, o-3). | Empty | azure_apim_gpt_deployment |
| Azure APIM Subscription Key | Provides the secret credential used when the selected authentication mode requires one. | Empty | azure_apim_gpt_subscription_key |
Embeddings
Embeddings
The Embeddings section belongs to the Embeddings tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.
Settings
| Setting | What it does | Default | Notes |
|---|---|---|---|
| Use APIM instead of direct to Azure OpenAI endpoint | Exposes the capability after required services, permissions, and rollout policy are ready. | Off | enable_embedding_apim; capability toggle |
| Azure OpenAI Embedding Endpoint | Provides the endpoint or route SimpleChat uses for this service. | Empty | azure_openai_embedding_endpoint |
| Authentication Type | Chooses whether SimpleChat authenticates to this service with a key, managed identity, or another supported method. | key | azure_openai_embedding_authentication_type |
| Subscription ID | Defines behavior for the related admin workflow; verify the affected feature after saving. | Empty | azure_openai_embedding_subscription_id |
| Resource Group | Defines behavior for the related admin workflow; verify the affected feature after saving. | Empty | azure_openai_embedding_resource_group |
| Azure OpenAI Embedding Key | Provides the secret credential used when the selected authentication mode requires one. | Empty | azure_openai_embedding_key |
| Azure OpenAI Embedding API Version | Pins the service API version SimpleChat sends with requests for this feature. | 2024-05-01-preview | azure_openai_embedding_api_version |
| Azure APIM Endpoint | Provides the endpoint or route SimpleChat uses for this service. | Empty | azure_apim_embedding_endpoint |
| Azure APIM API Version | Pins the service API version SimpleChat sends with requests for this feature. | Empty | azure_apim_embedding_api_version |
| Azure APIM Deployment | Selects the deployment SimpleChat sends requests to for this capability. | Empty | azure_apim_embedding_deployment |
| Azure APIM Subscription Key | Provides the secret credential used when the selected authentication mode requires one. | Empty | azure_apim_embedding_subscription_key |
Image Generation
Image Generation
The Image Generation section belongs to the Image Generation tab. Use it with the adjacent settings in this group so related rollout, access, and operational choices stay aligned.
Settings
| Setting | What it does | Default | Notes |
|---|---|---|---|
| Enable Image Generation | Exposes the capability after required services, permissions, and rollout policy are ready. | Off | enable_image_generation; capability toggle |
| Use APIM instead of direct to Azure OpenAI endpoint. | Exposes the capability after required services, permissions, and rollout policy are ready. | Off | enable_image_gen_apim; capability toggle |
| Azure OpenAI Image Generation Endpoint | Provides the endpoint or route SimpleChat uses for this service. | Empty | azure_openai_image_gen_endpoint |
| Authentication Type | Chooses whether SimpleChat authenticates to this service with a key, managed identity, or another supported method. | key | azure_openai_image_gen_authentication_type |
| Subscription ID | Defines behavior for the related admin workflow; verify the affected feature after saving. | Empty | azure_openai_image_gen_subscription_id |
| Resource Group | Defines behavior for the related admin workflow; verify the affected feature after saving. | Empty | azure_openai_image_gen_resource_group |
| Azure OpenAI Image Generation Key | Provides the secret credential used when the selected authentication mode requires one. | Empty | azure_openai_image_gen_key |
| Azure OpenAI Image Gen API Version | Pins the service API version SimpleChat sends with requests for this feature. | 2025-04-01-preview | azure_openai_image_gen_api_version |
| Azure APIM Endpoint | Provides the endpoint or route SimpleChat uses for this service. | Empty | azure_apim_image_gen_endpoint |
| Azure APIM API Version | Pins the service API version SimpleChat sends with requests for this feature. | Empty | azure_apim_image_gen_api_version |
| Azure APIM Deployment | Selects the deployment SimpleChat sends requests to for this capability. | Empty | azure_apim_image_gen_deployment |
| Azure APIM Subscription Key | Provides the secret credential used when the selected authentication mode requires one. | Empty | azure_apim_image_gen_subscription_key |
Common tasks
- Route chat through approved endpoints. Enable multi-endpoint management when needed and run a chat through the intended deployment. Outcome to verify: Requests reach the approved route.
- Configure embeddings. Set endpoint, authentication, API version, and deployment, then index a small document. Outcome to verify: Indexing succeeds with the configured embedding route.
- Enable image generation. Enable the capability, set endpoint values, and generate a low-risk test image. Outcome to verify: The image tool returns output from the configured deployment.
Troubleshooting
| Symptom | Likely cause | Fix |
|---|---|---|
| Embeddings fail during indexing | Endpoint, deployment, API version, or authentication does not match the Azure resource. | Validate the embedding route with a small document before bulk indexing. |