Manage model metadata
Create, edit, and delete model metadata to manage the AI models of your providers in an AI Gateway instance. You can also configure Credits consumption, token limits, capabilities, and available paths, which makes the models available to Model API and route configurations.
Model metadata overview
Model metadata is a specific AI model instance that belongs to a provider. Each model metadata entry includes:
Basic Information: the provider, name, and information source URL.
Credits Information: the unit Credits price for input, output, and cached token consumption.
Capabilities and Performance: the maximum input, output, and total token limits, and the input and output modalities.
Capabilities: the features that the model supports, such as Function Calling and Streaming Output.
Available Paths: the API paths and protocol types supported by the model.
Prerequisites
You have an AI Gateway instance. For instructions, see Create a gateway instance.
At least one provider, either preconfigured or custom, is available in the current instance. If no provider is available, create a provider or enable a preconfigured provider. For instructions, see Manage providers.
Limits
When you manage model metadata, note the following limits:
A path must start with
/.Duplicate available paths within the same model metadata entry are not recommended.
If you specify a Credits consumption parameter, the value must be a number greater than or equal to 0 with no more than two decimal places.
You cannot create model metadata in an instance that has no available provider.
View the model metadata list
Log on to the AI Gateway console, select the region where the instance resides, and then click the target instance ID.
In the left-side navigation pane, click Model Metadata Management to open the model metadata page.
On the Model Metadata tab, which is displayed by default, view the models in the current instance.
The list includes the following columns:
| Column | Description |
| Model Name | The display name of the model. |
| Unit Price | The unit Credits price for input, output, and cached tokens. Unit: Credits/1M tokens. |
| Total Token Limit | The total token limit for the model. |
| Capabilities | The ratio of enabled capabilities to all available capabilities. Example: 5/15. |
| Provider | The name of the provider associated with the model. |
| Updated At | The time when the resource was last updated. |
| Actions | View, Edit, and Delete. |
Create model metadata
On the model metadata page, click Create Model Metadata.
In the creation panel that opens, configure the parameters in each collapsible section in order. The following sections describe these parameters.
Click OK to create the model. The new model metadata entry appears in the model metadata list.
After creation, you can view the status of the entry in the model metadata list. Send a test request to verify that the path and token configurations have taken effect.
Basic Information
Basic Information
The following parameters provide the basic information of the model. Provider and Model Name are required.
| Parameter | Description |
| Provider | Required. The drop-down list displays all available providers in the current instance, including preconfigured and custom providers. If no provider is available, a message prompts you to create a provider or enable a preconfigured provider. |
| Model Name | Required. The display name of the model. |
Credits Information
Credits Information
The following parameters are optional and control Credits consumption when the model is called.
| Parameter | Description |
| Input Token Consumption | Optional. Unit: Credits/1M tokens. Default value: 0. |
| Output Token Consumption | Optional. Unit: Credits/1M tokens. Default value: 0. |
| Cached Token Consumption | Optional. Unit: Credits/1M tokens. Default value: 0. |
Capabilities and Performance
Capabilities and Performance
The following parameters are optional and describe the capability boundaries of the model.
| Parameter | Description |
| Maximum Input Tokens | Optional. The maximum number of input tokens. If specified, the value must be an integer greater than or equal to 0. |
| Maximum Output Tokens | Optional. The maximum number of output tokens. If specified, the value must be an integer greater than or equal to 0. |
| Total Token Limit | Optional. The total token limit. If specified, the value must be an integer greater than or equal to 0. |
| Input Modalities | Optional. You can select multiple values. Valid values: text, image, audio, and video. |
| Output Modalities | Optional. You can select multiple values. Valid values: text, image, audio, and video. |
Capabilities
Capabilities
The following parameters are optional. Select the capabilities that the model supports. You can select multiple values.
| Parameter | Description |
| System Message | Supports System Prompt. |
| Structured Output | Supports Structured Output. |
| Function Calling | Supports Function Calling. |
| Tool Choice | Supports Tool Choice. |
| Prefix Completion | Supports Prefix Completion. |
| Prompt Cache | Supports Prompt Cache. |
| Computer Use | Supports Computer Use. |
| PDF Input | Supports PDF Input. |
| Vision | Supports Vision. |
| Image Embedding | Supports Image Embedding. |
| Streaming Output | Supports Streaming Output. |
| Parallel Function Calling | Supports Parallel Function Calling. |
| Web Search | Supports Web Search. |
| Reasoning | Supports Reasoning. |
| URL Context | Supports URL Context. |
Available Paths
Available Paths
The following parameters are required and specify the API paths and protocols that the model supports.
| Parameter | Description |
| Path | Required. An API path that the model supports. The path must start with /. Example: /v1/chat/completions. |
| Protocol Type | Required. Valid values: OpenAI Compatible, Anthropic Compatible, and Custom. |
To add an available path, click Add Path.
Edit model metadata
On the model metadata page, find the target model metadata entry and click Edit in the Actions column.
Modify the parameters that you need to adjust. You can modify the following parameters:
Basic Information: the provider and the model name.
Credits Information: the input, output, and cached token consumption.
Capabilities and Performance: the maximum input, output, and total token limits, and the input and output modalities.
Capabilities: the features that the model supports.
Available Paths: the path and the Protocol Type.
Click OK to save the changes.
Editing a model name affects the model name options that are shown in Model API and route configurations.
Delete model metadata
On the model metadata page, find the target model metadata entry and click Delete in the Actions column.
Confirm the deletion of the model.
The model metadata entry is removed from the model metadata list. Deleting the resource does not invalidate model names that are already stored as strings in Model API and route configurations.
What to do next
Associate the model that you created with a Model API configuration so that applications can call it.
Select the model as the destination model in a route configuration to distribute traffic.
To add more models, make sure that the required providers are available. For instructions, see Manage providers.