模型列表接口

返回体为列表结构，模型条目在 data 数组里，顶层另有 object、success 等标识字段。

每个模型条目包含以下字段：

The body is a list structure: model entries are in the data array, alongside top-level marker fields such as object and success.

Each model entry contains the following fields:

supported_endpoint_types 的常见取值与实际调用路径对应如下：

这只是常见取值，不是全部。Seedance 系列另有视频类取值（如与 video、Official 相关的项），具体以接口返回为准。

同一个模型可能同时支持多种格式。例如 Claude 系模型通常同时返回 anthropic 与 openai，既可以用原生格式调用，也可以走 OpenAI 兼容格式，按你的代码习惯选一种即可。

Common values of supported_endpoint_types and the actual call paths they correspond to:

These are common values only, not the complete set. The Seedance series has its own video-related values (such as items involving video and Official); go by what the endpoint actually returns.

One model may support several formats at once. Claude-family models, for example, usually return both anthropic and openai, so you can call them natively or via the OpenAI-compatible format — pick whichever fits your code.

三个维度配合看：

• 能不能调 —— 看 /v1/models 里有没有这个 id。列表只返回当前账号可用的模型，换 Key 或换分组结果会变
• 怎么调 —— 看 supported_endpoint_types。要 OpenAI 兼容就挑带 openai 的，要用原生格式就挑对应的取值
• 多少钱 —— 见模型价格与分组接口，按 model_ratio 结合所在分组换算

Look at three dimensions together:

• Can you call it — check whether /v1/models lists that id. The list returns only models available to the current account, and a different key or group changes the result
• How to call it — check supported_endpoint_types. For OpenAI-compatible, pick one with openai; for native format, pick the matching value
• What it costs — see Model pricing and groups, converting with model_ratio and your group

1. 列表范围由账号决定，不是全站模型。Key 的限制与分组不同，看到的清单也会不同
2. created 不反映真实上线时间，各模型返回的值基本相同，不能用来排序
3. 筛选在客户端做。清单一次返回全部可用模型，按 supported_endpoint_types 或模型名在本地过滤即可
4. 模型清单会变动，新增或下线不需要改代码，按 id 直接调用即可
5. 本接口只提供模型的身份信息。能力标签、价格、分组在模型价格与分组接口里，每个模型的请求体参数见各自接口页，比如对话模型看「通用对话接口」的请求参数一节

1. The list's scope depends on the account, not the whole site. Different key restrictions and groups see different lists
2. created doesn't reflect real launch timing — the values returned are essentially the same across models and can't be used for sorting
3. Filter client-side. The list returns every available model at once; filter locally by supported_endpoint_types or model name
4. The catalog changes. Models come and go without code changes; call them directly by id
5. This endpoint only provides model identity. Capability tags, prices, and groups are in Model pricing and groups; request body parameters per model are on each endpoint page — for chat models, see the request parameters section of "Chat completions"

错误响应的格式、状态码与重试建议见错误码与限流，常见报错的排查步骤见常见问题。

For error response formats, status codes, and retry guidance see Error codes and rate limits; for troubleshooting steps on common errors see the FAQ.

拉取全部可用模型

最简形式：一次拿全当前账号可用的模型。

只取模型 ID

组装模型选择器时通常只需要 ID 列表，用 jq 过滤即可。

curl -s --request GET \
  --url 'https://api.oairoute.com/v1/models' \
  --header 'Authorization: Bearer YOUR_OAIRoute_KEY' \
  | jq -r '.data[].id'

按调用方式筛模型

想在本地只保留支持原生 Claude 格式的模型，按 supported_endpoint_types 过滤。

curl -s --request GET \
  --url 'https://api.oairoute.com/v1/models' \
  --header 'Authorization: Bearer YOUR_OAIRoute_KEY' \
  | jq -r '.data[] | select(.supported_endpoint_types | index("anthropic")) | .id'

只保留图像生成模型

同理，筛出带 image-generation 的模型，用于图像接口的模型下拉。

curl -s --request GET \
  --url 'https://api.oairoute.com/v1/models' \
  --header 'Authorization: Bearer YOUR_OAIRoute_KEY' \
  | jq -r '.data[] | select(.supported_endpoint_types | index("image-generation")) | .id'

List all available models

The simplest form: fetch every model available to the current account in one call.

Take only model IDs

When building a model picker you usually need just the ID list; filter with jq.

curl -s --request GET \
  --url 'https://api.oairoute.com/v1/models' \
  --header 'Authorization: Bearer YOUR_OAIRoute_KEY' \
  | jq -r '.data[].id'

Filter by call style

To keep only models supporting native Claude format locally, filter on supported_endpoint_types.

curl -s --request GET \
  --url 'https://api.oairoute.com/v1/models' \
  --header 'Authorization: Bearer YOUR_OAIRoute_KEY' \
  | jq -r '.data[] | select(.supported_endpoint_types | index("anthropic")) | .id'

Keep only image-generation models

Similarly, filter for models with image-generation for an image endpoint's model dropdown.

curl -s --request GET \
  --url 'https://api.oairoute.com/v1/models' \
  --header 'Authorization: Bearer YOUR_OAIRoute_KEY' \
  | jq -r '.data[] | select(.supported_endpoint_types | index("image-generation")) | .id'

真实返回，截取两条：

{
  "data": [
    {
      "id": "gpt-5-nano",
      "object": "model",
      "created": 1626777600,
      "owned_by": "openai",
      "supported_endpoint_types": ["openai"]
    },
    {
      "id": "claude-haiku-4-5-20251001",
      "object": "model",
      "created": 1626777600,
      "owned_by": "claude",
      "supported_endpoint_types": ["anthropic", "openai"]
    }
  ],
  "object": "list",
  "success": true
}

A real response, trimmed to two entries:

{
  "data": [
    {
      "id": "gpt-5-nano",
      "object": "model",
      "created": 1626777600,
      "owned_by": "openai",
      "supported_endpoint_types": ["openai"]
    },
    {
      "id": "claude-haiku-4-5-20251001",
      "object": "model",
      "created": 1626777600,
      "owned_by": "claude",
      "supported_endpoint_types": ["anthropic", "openai"]
    }
  ],
  "object": "list",
  "success": true
}