Resources / ModelProvider
ModelProvider
Who makes models (Anthropic, OpenAI, Google, DeepSeek, Meta, Mistral, ...) and the catalog of what they make: families, capabilities, context windows, modalities, published prices and deprecation dates. It holds no credentials. The platform keeps a shared catalog; a project may add its own. The status adds the models that InferenceProviders report they can serve.
Overview
| API | intelligence.etorm.com/v1alpha1 |
| Resource | modelproviders (cluster-scoped inside its cluster) |
| Short names | modelprov |
| Available in | the platform and project clusters |
| Status | written by a controller through /status |
Minimal example
apiVersion: intelligence.etorm.com/v1alpha1
kind: ModelProvider
metadata:
name: example
spec: {}
Spec
| Field | Type | Required | Default | Allowed | Description |
displayName | string | | | up to 128 characters | The vendor's name. |
website | string | | | up to 256 characters | The vendor's public address (https). |
matchPrefixes | list of string | | | up to 20 items | Model id prefixes that belong to this vendor, used to attach the models an InferenceProvider lists (for example claude, gpt, gemini). |
models | list of object | | | up to 500 items | The curated catalog. |
models[].id | string | yes | | length 1 to 128 | The model's id as the vendor names it. |
models[].family | string | | | up to 64 characters | The family, for example claude-opus. |
models[].displayName | string | | | up to 128 characters | A readable name. |
models[].kind | string (one of) | | "chat" | chat, embedding | What it does. |
models[].contextWindow | integer | | | at least 1 | Input tokens it accepts. |
models[].maxOutputTokens | integer | | | at least 1 | Output tokens it can answer with. |
models[].modalities | list of string | | | up to 8 items | What it accepts. |
models[].capabilities | list of string | | | up to 16 items | For example tools, streaming, json, reasoning. |
models[].pricing | object | | | | The vendor's published price, only when it is published. |
models[].pricing.inputMicrosPerMillion | integer | | | at least 0 | Millionths of a USD per million input tokens. |
models[].pricing.outputMicrosPerMillion | integer | | | at least 0 | Millionths of a USD per million output tokens. |
models[].deprecatedOn | string | | | up to 10 characters | The date it is withdrawn (YYYY-MM-DD). |
models[].replacedBy | string | | | up to 128 characters | The model that replaces it. |
Status
| Field | Type | Required | Default | Allowed | Description |
phase | string | | | | Ready. |
modelCount | integer | | | | Models in the catalog, curated and discovered. |
models | list of object | | | | The curated models plus those InferenceProviders report; each with who serves it and whether it is deprecated. |
models[].id | string | | | | The model's id. |
models[].kind | string | | | | chat or embedding. |
models[].contextWindow | integer | | | | Input tokens. |
models[].curated | boolean | | | | Written in the spec (otherwise discovered). |
models[].deprecated | boolean | | | | Past its deprecation date. |
models[].servedBy | list of string | | | | InferenceProviders that list it, as project/name. |
conditions | list of object | | | | Detailed state. |
conditions[].type | string | | | | Aspect. |
conditions[].status | string | | | | True or False. |
conditions[].reason | string | | | | Machine-readable reason. |
conditions[].message | string | | | | Human-readable detail. |
Columns in listings
| Column | From |
|---|
| Vendor | .spec.displayName |
| Models | .status.modelCount |
| Phase | .status.phase |
All resources