--- title: "Get endpoint analytics" method: GET path: "/v2/projects/{projectId}/endpoints/{id}/analytics" tags: ["EndpointService"] --- # Get endpoint analytics `GET /v2/projects/{projectId}/endpoints/{id}/analytics` ## Request - Base URL: `https://api.together.ai` - URL: `https://api.together.ai/v2/projects/{projectId}/endpoints/{id}/analytics` - Auth: HTTP bearer Returns aggregated request, token, latency, throughput, error, and resource-utilization metrics for an endpoint over a time range. Optionally includes time-series buckets and a per-deployment breakdown. ## Path parameters - `projectId` string, required: Project identifier. - `id` string, required: Endpoint identifier. ## Query parameters - `deploymentId` string: Restrict to a single deployment under this endpoint. - `startTime` string, date-time: Inclusive start of the time range. Defaults to 24 hours ago if unset. - `endTime` string, date-time: Exclusive end of the time range. Defaults to now if unset. - `includeTimeSeries` boolean: When true, include per-bucket time series in the response. - `granularity` string: Time-series bucket duration, such as `1m`, `1h`, or `1d`. Defaults to `1d`. ## Response `200` OK - DEAnalyticsData: Endpoint-wide usage and performance analytics with optional time-series and per-deployment breakdowns. - `endpointId` string: ID of the endpoint summarized by these analytics. - `timeRange` DEMetricsTimeRange: Closed-open time range used by metrics and analytics responses. - `startTime` string, date-time: Inclusive start of the time range. - `endTime` string, date-time: Exclusive end of the time range. - `metrics` DEEndpointMetrics: Operational metrics aggregated across all deployments receiving traffic for an endpoint. - `endpointId` string: The endpoint these metrics describe. - `timeRange` DEMetricsTimeRange: Closed-open time range used by metrics and analytics responses. - `startTime` string, date-time: Inclusive start of the time range. - `endTime` string, date-time: Exclusive end of the time range. - `requestMetrics` DERequestMetrics: Request counts, rate, and status-code distribution over a time range. - `totalRequests` string: Total requests received during the time range. - `successfulRequests` string: Requests completed successfully during the time range. - `failedRequests` string: Requests that failed during the time range. - `requestsPerSecond` number: Average requests per second over the time range. - `requestsByStatusCode` object: Request counts keyed by HTTP status code. - `latencyMetrics` DELatencyMetrics: Time-to-first-token, end-to-end, and inter-token latency percentiles in milliseconds. - `ttftP50Ms` number: 50th-percentile time to first token, in milliseconds. - `ttftP90Ms` number: 90th-percentile time to first token, in milliseconds. - `ttftP99Ms` number: 99th-percentile time to first token, in milliseconds. - `latencyP50Ms` number: 50th-percentile end-to-end request latency, in milliseconds. - `latencyP90Ms` number: 90th-percentile end-to-end request latency, in milliseconds. - `latencyP99Ms` number: 99th-percentile end-to-end request latency, in milliseconds. - `itlP50Ms` number: 50th-percentile inter-token latency, in milliseconds. - `itlP90Ms` number: 90th-percentile inter-token latency, in milliseconds. - `itlP99Ms` number: 99th-percentile inter-token latency, in milliseconds. - `throughputMetrics` DEThroughputMetrics: Token, request, and batching throughput over a time range. - `tokensPerSecond` number: Average generated tokens per second. - `requestsPerSecond` number: Average completed requests per second. - `avgBatchSize` number: Average number of requests processed in each runtime batch. - `avgBatchDepth` number: Average number of batches queued or in flight in the serving engine. - `errorMetrics` DEErrorMetrics: Error rate and aggregate counts by error type. Individual error samples are not included. - `errorRate` number: Percentage in [0, 100]. - `errorsByType` object: Counts of errors keyed by error type (e.g. HTTP status code or error kind). - `resourceUtilization` DEResourceUtilization: Average compute, memory, and network utilization for replicas over a time range. - `cpuUtilization` number: Average CPU utilization across replicas, as a percentage. - `gpuUtilization` number: Average GPU compute utilization across replicas, as a percentage. - `memoryUtilization` number: Average system memory utilization across replicas, as a percentage. - `gpuMemoryUtilization` number: Average GPU memory utilization across replicas, as a percentage. - `networkBandwidthMbps` number: Average network throughput across replicas, in megabits per second. - `tokenMetrics` DETokenMetrics: Aggregate and per-request token usage over a time range. - `totalInputTokens` string: Total input tokens processed during the time range. - `totalOutputTokens` string: Total output tokens generated during the time range. - `avgInputTokens` number: Average input tokens per request. - `avgOutputTokens` number: Average output tokens per request. - `deploymentMetrics` DEDeploymentMetrics[]: Per-deployment breakdown, if the endpoint has multiple deployments. - `deploymentId` string: ID of the deployment summarized by these metrics. - `endpointId` string: ID of the deployment's parent endpoint. - `timeRange` DEMetricsTimeRange: Closed-open time range used by metrics and analytics responses. - `startTime` string, date-time: Inclusive start of the time range. - `endTime` string, date-time: Exclusive end of the time range. - `requestMetrics` DERequestMetrics: Request counts, rate, and status-code distribution over a time range. - `totalRequests` string: Total requests received during the time range. - `successfulRequests` string: Requests completed successfully during the time range. - `failedRequests` string: Requests that failed during the time range. - `requestsPerSecond` number: Average requests per second over the time range. - `requestsByStatusCode` object: Request counts keyed by HTTP status code. - `latencyMetrics` DELatencyMetrics: Time-to-first-token, end-to-end, and inter-token latency percentiles in milliseconds. - `ttftP50Ms` number: 50th-percentile time to first token, in milliseconds. - `ttftP90Ms` number: 90th-percentile time to first token, in milliseconds. - `ttftP99Ms` number: 99th-percentile time to first token, in milliseconds. - `latencyP50Ms` number: 50th-percentile end-to-end request latency, in milliseconds. - `latencyP90Ms` number: 90th-percentile end-to-end request latency, in milliseconds. - `latencyP99Ms` number: 99th-percentile end-to-end request latency, in milliseconds. - `itlP50Ms` number: 50th-percentile inter-token latency, in milliseconds. - `itlP90Ms` number: 90th-percentile inter-token latency, in milliseconds. - `itlP99Ms` number: 99th-percentile inter-token latency, in milliseconds. - `throughputMetrics` DEThroughputMetrics: Token, request, and batching throughput over a time range. - `tokensPerSecond` number: Average generated tokens per second. - `requestsPerSecond` number: Average completed requests per second. - `avgBatchSize` number: Average number of requests processed in each runtime batch. - `avgBatchDepth` number: Average number of batches queued or in flight in the serving engine. - `errorMetrics` DEErrorMetrics: Error rate and aggregate counts by error type. Individual error samples are not included. - `errorRate` number: Percentage in [0, 100]. - `errorsByType` object: Counts of errors keyed by error type (e.g. HTTP status code or error kind). - `resourceUtilization` DEResourceUtilization: Average compute, memory, and network utilization for replicas over a time range. - `cpuUtilization` number: Average CPU utilization across replicas, as a percentage. - `gpuUtilization` number: Average GPU compute utilization across replicas, as a percentage. - `memoryUtilization` number: Average system memory utilization across replicas, as a percentage. - `gpuMemoryUtilization` number: Average GPU memory utilization across replicas, as a percentage. - `networkBandwidthMbps` number: Average network throughput across replicas, in megabits per second. - `tokenMetrics` DETokenMetrics: Aggregate and per-request token usage over a time range. - `totalInputTokens` string: Total input tokens processed during the time range. - `totalOutputTokens` string: Total output tokens generated during the time range. - `avgInputTokens` number: Average input tokens per request. - `avgOutputTokens` number: Average output tokens per request. - `timeSeries` DETimeSeriesDataPoint[]: Per-bucket metric samples, included only when `includeTimeSeries` is true. - `timestamp` string, date-time: Start time of the metric bucket. - `values` object: Metric names mapped to their numeric values for this bucket. - `deploymentAnalytics` DEDeploymentAnalyticsData[]: Per-deployment analytics. - `deploymentId` string: ID of the deployment summarized by these analytics. - `endpointId` string: ID of the deployment's parent endpoint. - `timeRange` DEMetricsTimeRange: Closed-open time range used by metrics and analytics responses. - `startTime` string, date-time: Inclusive start of the time range. - `endTime` string, date-time: Exclusive end of the time range. - `metrics` DEDeploymentMetrics: Operational metrics for one deployment under an endpoint. - `deploymentId` string: ID of the deployment summarized by these metrics. - `endpointId` string: ID of the deployment's parent endpoint. - `timeRange` DEMetricsTimeRange: Closed-open time range used by metrics and analytics responses. - `startTime` string, date-time: Inclusive start of the time range. - `endTime` string, date-time: Exclusive end of the time range. - `requestMetrics` DERequestMetrics: Request counts, rate, and status-code distribution over a time range. - `totalRequests` string: Total requests received during the time range. - `successfulRequests` string: Requests completed successfully during the time range. - `failedRequests` string: Requests that failed during the time range. - `requestsPerSecond` number: Average requests per second over the time range. - `requestsByStatusCode` object: Request counts keyed by HTTP status code. - `latencyMetrics` DELatencyMetrics: Time-to-first-token, end-to-end, and inter-token latency percentiles in milliseconds. - `ttftP50Ms` number: 50th-percentile time to first token, in milliseconds. - `ttftP90Ms` number: 90th-percentile time to first token, in milliseconds. - `ttftP99Ms` number: 99th-percentile time to first token, in milliseconds. - `latencyP50Ms` number: 50th-percentile end-to-end request latency, in milliseconds. - `latencyP90Ms` number: 90th-percentile end-to-end request latency, in milliseconds. - `latencyP99Ms` number: 99th-percentile end-to-end request latency, in milliseconds. - `itlP50Ms` number: 50th-percentile inter-token latency, in milliseconds. - `itlP90Ms` number: 90th-percentile inter-token latency, in milliseconds. - `itlP99Ms` number: 99th-percentile inter-token latency, in milliseconds. - `throughputMetrics` DEThroughputMetrics: Token, request, and batching throughput over a time range. - `tokensPerSecond` number: Average generated tokens per second. - `requestsPerSecond` number: Average completed requests per second. - `avgBatchSize` number: Average number of requests processed in each runtime batch. - `avgBatchDepth` number: Average number of batches queued or in flight in the serving engine. - `errorMetrics` DEErrorMetrics: Error rate and aggregate counts by error type. Individual error samples are not included. - `errorRate` number: Percentage in [0, 100]. - `errorsByType` object: Counts of errors keyed by error type (e.g. HTTP status code or error kind). - `resourceUtilization` DEResourceUtilization: Average compute, memory, and network utilization for replicas over a time range. - `cpuUtilization` number: Average CPU utilization across replicas, as a percentage. - `gpuUtilization` number: Average GPU compute utilization across replicas, as a percentage. - `memoryUtilization` number: Average system memory utilization across replicas, as a percentage. - `gpuMemoryUtilization` number: Average GPU memory utilization across replicas, as a percentage. - `networkBandwidthMbps` number: Average network throughput across replicas, in megabits per second. - `tokenMetrics` DETokenMetrics: Aggregate and per-request token usage over a time range. - `totalInputTokens` string: Total input tokens processed during the time range. - `totalOutputTokens` string: Total output tokens generated during the time range. - `avgInputTokens` number: Average input tokens per request. - `avgOutputTokens` number: Average output tokens per request. - `timeSeries` DETimeSeriesDataPoint[]: Per-bucket metric samples for the deployment. - `timestamp` string, date-time: Start time of the metric bucket. - `values` object: Metric names mapped to their numeric values for this bucket. ## Other responses - `default`: Default error response ## Changes - **2026-10-22** `bb5257b3173a`: 1 info - endpoint added [Operation history](https://skmtc.dev/together/apis/together-apis/changes/v2/projects/:projectId/endpoints/:id/analytics/get.md) --- [API](https://skmtc.dev/together/apis/together-apis.md) - [All operations](https://skmtc.dev/together/apis/together-apis/llms.txt) - [OpenAPI document](https://skmtc.dev/together/apis/together-apis/revisions/93438ddac711?raw)