Required. Unique name for this configuration. Use only letters, numbers, hyphens, and underscores (for example gpt4-eastus-prod).

Azure OpenAI Service. The provider for this saved configuration.

Optional for this provider when Deployment name identifies the model deployment you call. Use this field only if your integration still expects a model identifier here; otherwise leave blank or follow your administrator’s convention. On other providers, this field is typically required.

Optional. Leave blank to supply the key at runtime. If you enter a key, it is stored with this configuration—treat it as a secret and follow your organization’s access and rotation rules. Use the key from your Azure OpenAI resource (or managed identity / key vault flow if your product supports it).

Required. The endpoint URL for your Azure OpenAI resource—usually https://<resource-name>.openai.azure.com/ or the base path your gateway uses. Must match the resource where the deployment lives.

Required. Name of the model deployment in Azure OpenAI (as shown in Azure AI Foundry or Azure portal). Requests are routed to this deployment; it must exist and match the model you intend to use.

Optional in some setups; required when your client must pin a REST api-version query parameter. Use a value supported for your operation (for example 2024-02-15-preview). See Microsoft’s API version notes for your scenario.

Optional. Service or feature version string when your integration distinguishes API or service revisions beyond API version. Use only if your product documentation requires it; otherwise leave blank for default behavior.

Leave blank to use the default. When set, use 0 or greater. Higher values increase randomness; many models use a typical range around 0–2. Exact limits depend on the model and API version.

Leave blank to use the default. When set, enter a positive integer for the maximum number of tokens in the completion (per Azure OpenAI request limits for your model).

Leave blank to use the default. When set, use 0 or greater (often 0–1). Nucleus sampling; may be ignored if unsupported for the deployment.

Leave blank to use the default. When set, use 0 or greater. Reduces likelihood of repeating topics already present in the context.

Leave blank to use the default. When set, use values allowed by Azure OpenAI for your API version (often in the -2.0–2.0 range). Reduces repetition of the same tokens. Shown only if your UI exposes this control.

Leave blank to use the default. When set, use a positive integer when the service supports deterministic sampling for your model; behavior may still vary by model or version.

Default or structured output (for example JSON) when your integration and model support it. Leave Default unless you rely on structured responses.

Optional. Comma-separated sequences that stop generation (for example END or newline tokens your API accepts).

Optional. HTTP proxy hostname or IP. If you set a host, you must set Proxy port.

Optional. Integer from 1 through 65535. Required when proxy host is set.

Optional. User name for proxy authentication. Leave blank if the proxy does not require credentials.

Optional. Proxy password. Stored credentials are sensitive—follow your organization’s secret-handling rules. Leave blank if not required.

Leave blank to use the default. When set, enter a positive integer for how many times to retry a failed request to the Azure endpoint.

Leave blank to use the default. When set, enter a positive integer for the maximum wait time per request (including large completions).

Optional. End-user identifier for abuse monitoring, attribution, or policy (maximum 256 characters) when your integration passes it to the service. Leave blank if not used.

When enabled, request details are written to logs. Requests can include prompts, headers, and Azure resource metadata—use only while troubleshooting and disable afterward. Follow your organization’s logging and privacy requirements.

When enabled, response details are written to logs. Responses can contain generated text and personal data—use only while troubleshooting and disable afterward. Follow your organization’s logging and privacy requirements.