List models — HopsAPI Docs

GET/v1/models

Returns every model available to the current API key, including channel-aware LLM discovery plus protocol and endpoint metadata.

curl https://hopsapi.com/v1/models \
  -H "Authorization: Bearer sk-YOUR_KEY"

Each item includes:

Field	Meaning
`id`	Model ID to send in requests
`type`	`llm`, `image`, or `media`. `image` is a text model that still uses `generateContent` and has an image output price. A Gemini image model published as media is `media` and uses `/v1/router/task/create`
`supported_protocols`	Every protocol ID enabled for the model
`protocol_endpoints`	Every supported protocol mapped to its endpoint and request format
`format`	Compatibility field for the primary route: `openai`, `anthropic`, `gemini`, or `task`
`endpoint`	Compatibility field for the model's primary endpoint
`responses_endpoint`	Compatibility field; present as `/v1/responses` when `supported_protocols` includes `openai-responses`

A Responses-capable item may look like:

{
  "id": "gpt-5.6-sol",
  "type": "llm",
  "format": "openai",
  "endpoint": "/v1/chat/completions",
  "supported_protocols": ["openai-completions", "openai-responses"],
  "protocol_endpoints": [
    {
      "protocol": "openai-completions",
      "endpoint": "/v1/chat/completions",
      "format": "openai"
    },
    {
      "protocol": "openai-responses",
      "endpoint": "/v1/responses",
      "format": "openai-responses"
    }
  ],
  "responses_endpoint": "/v1/responses"
}

Use each model ID exactly as returned. For a model-configured API key, this endpoint lists only models authorized for that key and still available under the account's current permissions and upstream routes. An LLM ID is routed through the channel selected for that model; different models on one key can use different channels. Media models keep their existing IDs and normally use Standard; GPT Image 2 is the exception and supports Standard or Flex. Multi-protocol clients should prefer supported_protocols and protocol_endpoints; format, endpoint, and responses_endpoint remain for backward compatibility.

supported_protocols describes the route behind the channel this key uses for that model, so one model can expose different protocols to two keys authorized on different channels.

Pass ?protocol= to list only the models callable with a single protocol. It accepts a protocol ID (openai-completions, openai-responses, anthropic-messages, google-generative-ai) or a short alias (openai, responses, anthropic, gemini); an unknown value returns 400. Without the parameter, an anthropic-version or x-goog-api-key request header selects the matching protocol. With none of these, the full list is returned. x-goog-api-key and ?key= also authenticate this endpoint. ?key= does not select a protocol by itself.

The Claude 5 models enabled by default use the IDs claude-fable-5 and claude-sonnet-5. Their primary format is anthropic and their primary endpoint is /v1/messages. Responses availability is determined by whether supported_protocols includes openai-responses (or, for older clients, whether responses_endpoint is present), rather than by the model name alone.

For gemini-3.5-flash, supported_protocols includes both openai-completions and google-generative-ai, mapped to /v1/chat/completions and /v1/models/{model}:generateContent. OpenAI compatibility for any other Gemini model is determined by that model's returned capabilities.

claude-opus-5 is preregistered for Standard and Flex with format="anthropic", endpoint="/v1/messages", and an initial $5/$25 input/output price per million tokens. It remains hidden by default and is callable only when GET /v1/models returns that ID for the current API key.

claude-opus-5-5 is available on Official API Relay (standard) only, with format="anthropic" and endpoint="/v1/messages". Subscription Direct (flex) keys do not list this ID. Official prices are $4 input and $20 output per million tokens; cache read is $0.20, 5-minute cache write is $5, and 1-hour cache write is $8. There is no long-context surcharge. Call it only when GET /v1/models returns the ID for the current API key. See Claude Opus 5.5.

claude-sonnet-5-5 is available on Subscription Direct (flex) only, with format="anthropic" and endpoint="/v1/messages". Official API Relay (standard) keys do not list this ID. Official prices are $2 input and $10 output per million tokens; cache read is $0.20, 5-minute cache write is $2.50, and 1-hour cache write is $4. There is no long-context surcharge. Call it only when GET /v1/models returns the ID for the current API key. See Claude Sonnet 5.5.

gpt-6.1-sol uses OpenAI Chat Completions, and OpenAI Responses when supported_protocols includes openai-responses. Official prices per million tokens are $2 input, $0.10 cached input, and $10 output. Prompts with more than 400,000 input tokens use $4 input, $0.20 cached input, and $15 output for that whole request. Cache writes are not billed. Call it only when GET /v1/models returns the ID for the current API key. See GPT-6.1 Sol.

On this page