Models

GET /v1/models
GET /v1/models/:model

GET /v1/models lists every model the gateway will advertise for this key: platform providers, org-scoped custom deployments, and BYOK adapters. Rows are enriched with catalog status and list prices when those exist in Postgres.

bash
export OPENDOOR_API_KEY=opd_…
export OPENDOOR_BASE_URL=https://opendoor-gcp.web.app

curl "$OPENDOOR_BASE_URL/v1/models" \
  -H "Authorization: Bearer $OPENDOOR_API_KEY"

json
{
  "object": "list",
  "data": [
    {
      "id": "gemma-4-26b-a4b-it",
      "object": "model",
      "owned_by": "google",
      "provider": "vertex",
      "deployment_status": "live",
      "display_name": "Gemma 4 26B",
      "context_window": 128000,
      "supports_vision": false,
      "supports_tools": true,
      "supports_json_mode": true,
      "pricing": {
        "prompt": 0.15,
        "completion": 0.6
      },
      "architecture": {
        "modality": "text",
        "input_modalities": ["text"],
        "output_modalities": ["text"]
      },
      "supported_parameters": ["temperature", "max_tokens", "top_p", "stream", "tools", "tool_choice"],
      "top_provider": { "slug": "vertex" },
      "serverless": true
    }
  ]
}

Pricing numbers are catalog list rates (per 1M tokens in prompt / completion when present). Always treat GET /v1/models as source of truth for your workspace.

Hosted Vertex MaaS ids

These catalog ids are the serverless wholesale set the gateway advertises when Vertex (or Together overflow) is configured:

gemma-4-26b-a4b-it, qwen3-next-80b-instruct, qwen3-next-80b-thinking, qwen3-coder-480b-a35b-instruct, deepseek-v3.2, deepseek-r1, kimi-k2-thinking, minimax-m2, glm-4.7, glm-5, gpt-oss-120b, gpt-oss-20b.

Image / video ids (when Vertex media is configured): gemini-2.5-flash-image, gemini-3.1-flash-image, gemini-3-pro-image, veo-3.1-fast-generate-001, veo-3.1-generate-001.

Deployment status

Status	Meaning	Chat
`live`	Callable now	200
`warming`	Listed; not serving yet	503
`dedicated`	Needs a GPU from Dashboard → Deployments	400
`available_on_request`	Closed / marketplace only (open-weight is not blocked on this flag)	400 for closed
`coming_soon`	Not callable yet	400

Filter to live models:

bash
curl "$OPENDOOR_BASE_URL/v1/models" \
  -H "Authorization: Bearer $OPENDOOR_API_KEY" \
  | jq '.data | map(select(.deployment_status == "live"))'

One model

bash
curl "$OPENDOOR_BASE_URL/v1/models/gemma-4-26b-a4b-it" \
  -H "Authorization: Bearer $OPENDOOR_API_KEY"

Returns { id, object, created, owned_by } for a routed id, or 404 { "error": "Model not found" }.

SDKs

ts
const models = await client.models.list();

python
print(client.models.list())

bash
bun run od -- models

Chat

Errors