NewAPI Compatibility

The main Inference Space APIs are OpenAI- and Anthropic-compatible. If your system was built on NewAPI (or the wider OneAPI ecosystem), you can keep using the endpoints below to look up models, balance, request logs, and daily consumption. They are all read-only.

Authentication

All endpoints take a gk_ API Key in either header:

Authorization: Bearer gk_xxxxxxxxxxxxxx
x-api-key: gk_xxxxxxxxxxxxxx

Authentication failures return a Chinese-language message with error.code set to missing_api_key, invalid_api_key, insufficient_scope, or ip_not_allowed:

{
  "success": false,
  "message": "API Key 无效或已停用",
  "error": { "code": "invalid_api_key", "message": "API Key 无效或已停用" }
}

Model Discovery

List the models the current Key can call:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/v1/models" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY"

{
  "object": "list",
  "data": [
    { "id": "gpt-5.6-sol", "object": "model", "created": 0, "owned_by": "wujie" },
    { "id": "gpt-image-2", "object": "model", "created": 0, "owned_by": "wujie-image" }
  ]
}

• Results are filtered by the Key's permissions: chat models need ai:llm, image models need ai:image, and video models need ai:video. If the Key has an allowed-model list, only those models are returned.
• owned_by is the model category: wujie (chat), wujie-image (image), wujie-video (video).

Look up a single model:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/v1/models/gpt-image-2" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY"

If the model does not exist or the Key cannot access it, the response is 404:

{
  "success": false,
  "message": "模型不存在或当前 API Key 无权访问",
  "error": { "code": "model_not_found", "message": "模型不存在或当前 API Key 无权访问" }
}

Gemini format

Gemini SDKs and native clients can use /v1beta/models for a Gemini-format list. It returns only Gemini-family models such as Gemini, Imagen, Nano Banana, and Veo:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/v1beta/models" \
  -H "x-goog-api-key: $INFERENCE_SPACE_API_KEY"

{
  "models": [
    {
      "name": "models/gemini-3-pro-image",
      "displayName": "gemini-3-pro-image",
      "version": "001",
      "supportedGenerationMethods": ["generateContent", "streamGenerateContent"]
    }
  ]
}

• Besides Authorization / x-api-key, this endpoint also accepts Gemini's x-goog-api-key header and a ?key= query parameter. Query strings are easily logged, so prefer a header.
• Video models report supportedGenerationMethods: ["predictLongRunning"].
• /v1beta/openai/models returns the same OpenAI-format list as /v1/models (all categories) and does not accept ?key=.

OpenAI Audio Aliases

translations uses the same speech recognition as transcriptions and does not translate. speech accepts both OpenAI fields input / response_format and the native fields text / format. See Speech recognition and synthesis.

Balance Lookup

Get the wallet balance and total spend of the Key's organization:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/api/usage/token/" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY"

{
  "code": true,
  "message": "ok",
  "data": {
    "object": "token_usage",
    "name": "Production service",
    "total_granted": 191.34,
    "total_used": 67.89,
    "total_available": 123.45,
    "unlimited_quota": false,
    "model_limits": {},
    "model_limits_enabled": false,
    "expires_at": 0
  }
}

All Keys in an organization share one wallet and see the same balance. A Key has no quota of its own; it only determines permissions and which Key a log entry belongs to.

OpenAI Dashboard Balance Aliases

Dashboard Billing lookups for older clients, with or without the /v1 prefix:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/v1/dashboard/billing/subscription" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY"

subscription returns:

{
  "object": "billing_subscription",
  "has_payment_method": true,
  "soft_limit_usd": 123.45,
  "hard_limit_usd": 123.45,
  "system_hard_limit_usd": 123.45,
  "access_until": 0
}

• soft_limit_usd is the current balance. When the organization has a credit limit, hard_limit_usd / system_hard_limit_usd are that limit; otherwise they equal the current balance.
• usd in the field names is kept only for compatibility; amounts are in the organization's billing currency.

usage returns {"object": "list", "total_usage": 6789}. total_usage is the organization's total spend in the smallest currency unit (cents, or fen for CNY billing).

Request Log Lookup

Query the request logs of the Key in the request header:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/api/log/token?page_size=20&start_timestamp=1783076400&end_timestamp=1783681200" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY"

{
  "success": true,
  "message": "",
  "data": [
    {
      "id": "call_...",
      "created_at": 1783581600,
      "type": "llm",
      "model_name": "gpt-5.6-sol",
      "token_name": "Production service",
      "token_id": "key_...",
      "quota": 0.12,
      "prompt_tokens": 7,
      "completion_tokens": 5,
      "use_time": 42,
      "is_stream": false,
      "request_id": "req_...",
      "ip": "203.0.113.10",
      "content": ""
    }
  ]
}

Notes:

• data: [] only means this Key has no calls in the range; the endpoint is working.
• Calls that finally returned 4xx / 5xx appear in the list with the error in content. If a request succeeded after retries, only the final success is recorded.
• Only paging and time-range filters are supported; there is no filter by status code or model. Use the request log page in the console for precise filtering.
• If log storage is temporarily unavailable, the endpoint returns 503 log_store_unavailable; retry later.

Daily Consumption Lookup

Daily spend totals for the Key's organization, for reconciling against the daily consumption records in the console:

# Global region; China region is https://cn.inf.space (accelerated) or https://global.inf.space (international)
curl "https://ai.inf.space/api/consumption/daily?start_timestamp=1783076400&end_timestamp=1783681200" \
  -H "Authorization: Bearer $INFERENCE_SPACE_API_KEY"

{
  "success": true,
  "message": "",
  "data": {
    "days": [
      {
        "date": "2026-07-01",
        "currency": "CNY",
        "log_net_amount": 12.34,
        "log_gross_amount": 12.34,
        "log_request_count": 256,
        "correction_amount": 0,
        "corrected_amount": 12.34,
        "corrections": []
      }
    ],
    "totals": {
      "CNY": { "log_net_amount": 12.34, "correction_amount": 0, "corrected_amount": 12.34, "log_request_count": 256 }
    }
  }
}

• log_net_amount: the day's charges summed from the request logs. log_request_count: number of calls that day.
• correction_amount: the day's reconciliation adjustments (0 if none); details are in corrections.
• corrected_amount: the day's spend after adjustments, log_net_amount + correction_amount.
• totals sums the whole range per currency.

Request ID Correlation

Every gateway response carries two headers with the same value, X-Gateway-Request-Id and X-Oneapi-Request-Id. NewAPI reads the latter as its upstream request ID. If your NewAPI forwards an X-Request-Id or X-Oneapi-Request-Id request header (1–128 letters, digits, or _ . : -), the gateway reuses that ID, so logs on both sides line up directly.

Unsupported Items

The following common NewAPI endpoints are not available yet. Integrate with the native Inference Space APIs instead: