Telemetry API

The Telemetry API provides real-time and historical observability data for inference providers. Use it to check provider health status, measure latency metrics, and retrieve providers ranked by a composite score combining uptime, latency, and price.

Provider Health


GET /api/telemetry/health/:provider_id

Returns health status history for a provider, including up/down status checks with timestamps and response times.

Path Parameters

Parameter	Type	Required	Description
`provider_id`	string	Yes	The provider’s unique identifier (e.g. `openai`, `anthropic`)

Query Parameters

Parameter	Type	Required	Default	Description
`period`	string	No	`7d`	Time period for health history (e.g. `1d`, `7d`, `30d`)
`limit`	integer	No	`50`	Maximum number of health check entries to return (max 500)

Example Request


curl "https://api.openmodels.run/api/telemetry/health/openai?period=7d&limit=20"

Example Response


{
  "provider_id": "openai",
  "provider_name": "OpenAI",
  "period": "7d",
  "uptime_percentage": 99.82,
  "checks": [
    {
      "status": "up",
      "checked_at": "2025-07-10T14:30:00.000Z",
      "response_time_ms": 142
    },
    {
      "status": "up",
      "checked_at": "2025-07-10T14:25:00.000Z",
      "response_time_ms": 138
    },
    {
      "status": "down",
      "checked_at": "2025-07-10T03:15:00.000Z",
      "response_time_ms": null
    },
    {
      "status": "up",
      "checked_at": "2025-07-10T03:10:00.000Z",
      "response_time_ms": 156
    }
  ]
}

Response Fields

Field	Type	Description
`provider_id`	string	The requested provider’s identifier
`provider_name`	string	Human-readable provider name
`period`	string	The requested time period
`uptime_percentage`	number	Percentage of successful health checks in the period (0–100)
`checks`	array	Array of health check entries, ordered by most recent first
`checks[].status`	string	Health check result: `up` or `down`
`checks[].checked_at`	string	ISO 8601 timestamp when the check was performed
`checks[].response_time_ms`	integer \| null	Response time in milliseconds, or `null` if the check failed

Error Responses

Status	Type	Description
`400`	`BAD_REQUEST`	Invalid query parameters (e.g. unsupported `period` value)
`404`	`NOT_FOUND`	Provider with the given ID does not exist

Provider Latency


GET /api/telemetry/latency/:provider_id

Returns latency metrics for a provider, including time-to-first-token and total response time measurements.

Path Parameters

Parameter	Type	Required	Description
`provider_id`	string	Yes	The provider’s unique identifier (e.g. `together-ai`, `deepseek`)

Query Parameters

Parameter	Type	Required	Default	Description
`period`	string	No	`7d`	Time period for latency data (e.g. `1d`, `7d`, `30d`)
`limit`	integer	No	`50`	Maximum number of latency measurements to return (max 500)

Example Request


curl "https://api.openmodels.run/api/telemetry/latency/together-ai?period=7d&limit=10"

Example Response


{
  "provider_id": "together-ai",
  "provider_name": "Together AI",
  "period": "7d",
  "average_ttft_ms": 245,
  "average_total_ms": 1820,
  "p95_ttft_ms": 410,
  "p95_total_ms": 3200,
  "measurements": [
    {
      "time_to_first_token_ms": 230,
      "total_response_time_ms": 1750,
      "measured_at": "2025-07-10T14:30:00.000Z"
    },
    {
      "time_to_first_token_ms": 255,
      "total_response_time_ms": 1890,
      "measured_at": "2025-07-10T14:25:00.000Z"
    },
    {
      "time_to_first_token_ms": 410,
      "total_response_time_ms": 3150,
      "measured_at": "2025-07-10T14:20:00.000Z"
    },
    {
      "time_to_first_token_ms": 198,
      "total_response_time_ms": 1620,
      "measured_at": "2025-07-10T14:15:00.000Z"
    }
  ]
}

Response Fields

Field	Type	Description
`provider_id`	string	The requested provider’s identifier
`provider_name`	string	Human-readable provider name
`period`	string	The requested time period
`average_ttft_ms`	integer	Average time-to-first-token in milliseconds over the period
`average_total_ms`	integer	Average total response time in milliseconds over the period
`p95_ttft_ms`	integer	95th percentile time-to-first-token in milliseconds
`p95_total_ms`	integer	95th percentile total response time in milliseconds
`measurements`	array	Array of latency measurements, ordered by most recent first
`measurements[].time_to_first_token_ms`	integer	Time to first token in milliseconds
`measurements[].total_response_time_ms`	integer	Total response time in milliseconds
`measurements[].measured_at`	string	ISO 8601 timestamp when the measurement was taken

Error Responses

Status	Type	Description
`400`	`BAD_REQUEST`	Invalid query parameters (e.g. unsupported `period` value)
`404`	`NOT_FOUND`	Provider with the given ID does not exist

Ranked Providers


GET /api/telemetry/ranked/:model_id

Returns providers ranked by a composite score for a given model. The score combines uptime reliability, latency performance, and pricing to help you choose the best provider.

Path Parameters

Parameter	Type	Required	Description
`model_id`	string	Yes	The model’s unique identifier (e.g. `deepseek-v3`, `gpt-5`)

Query Parameters

Parameter	Type	Required	Default	Description
`period`	string	No	`7d`	Time period for scoring data (e.g. `1d`, `7d`, `30d`)
`limit`	integer	No	`10`	Maximum number of ranked providers to return (max 50)

Example Request


curl "https://api.openmodels.run/api/telemetry/ranked/deepseek-v3?period=7d&limit=5"

Example Response


{
  "model_id": "deepseek-v3",
  "model_name": "DeepSeek V3",
  "period": "7d",
  "ranked_providers": [
    {
      "rank": 1,
      "provider_id": "deepseek",
      "provider_name": "DeepSeek",
      "composite_score": 94.5,
      "uptime_percentage": 99.9,
      "average_ttft_ms": 180,
      "average_total_ms": 1400,
      "input_per_million": 0.27,
      "output_per_million": 1.10,
      "currency": "USD"
    },
    {
      "rank": 2,
      "provider_id": "together-ai",
      "provider_name": "Together AI",
      "composite_score": 91.2,
      "uptime_percentage": 99.7,
      "average_ttft_ms": 245,
      "average_total_ms": 1820,
      "input_per_million": 0.50,
      "output_per_million": 0.90,
      "currency": "USD"
    },
    {
      "rank": 3,
      "provider_id": "deepinfra",
      "provider_name": "DeepInfra",
      "composite_score": 88.7,
      "uptime_percentage": 99.5,
      "average_ttft_ms": 290,
      "average_total_ms": 2100,
      "input_per_million": 0.35,
      "output_per_million": 0.80,
      "currency": "USD"
    }
  ]
}

Response Fields

Field	Type	Description
`model_id`	string	The requested model’s identifier
`model_name`	string	Human-readable model name
`period`	string	The time period used for scoring
`ranked_providers`	array	Array of providers sorted by composite score (descending)
`ranked_providers[].rank`	integer	Position in the ranking (1-indexed)
`ranked_providers[].provider_id`	string	Unique provider identifier
`ranked_providers[].provider_name`	string	Human-readable provider name
`ranked_providers[].composite_score`	number	Composite score from 0–100 combining uptime, latency, and price
`ranked_providers[].uptime_percentage`	number	Provider uptime percentage over the period (0–100)
`ranked_providers[].average_ttft_ms`	integer	Average time-to-first-token in milliseconds
`ranked_providers[].average_total_ms`	integer	Average total response time in milliseconds
`ranked_providers[].input_per_million`	number	Cost per 1M input tokens
`ranked_providers[].output_per_million`	number	Cost per 1M output tokens
`ranked_providers[].currency`	string	ISO 4217 currency code

Error Responses

Status	Type	Description
`400`	`BAD_REQUEST`	Invalid query parameters (e.g. unsupported `period` value)
`404`	`NOT_FOUND`	Model with the given ID does not exist or has no providers