📜 Logs and Async Tasks

View API requests, drawing, and async task records, filter details by criteria, and track status, results, and refunds.

The platform categorizes and stores API request details, drawing records, and long-running asynchronous tasks such as audio and video, making it easy for developers to quickly troubleshoot, reconcile accounts accurately, and trace results.

The console provides three dedicated log sections:

/usage-logs/common: General API call details and Token billing logs;
/usage-logs/drawing: Midjourney drawing task logs and image results;
/usage-logs/task: Logs for async generation tasks such as Suno music, videos, etc.

User logs are strictly isolated by permissions. Standard developers can only view request records and task results initiated by the current account.

Usage Logs

Click Usage Logs in the left sidebar or directly access /usage-logs/common to view API requests and billing details.

Filter Records

Filter Option	Description
Time Range	Filter by request creation time, supporting custom start and end dates.
Model Name	Enter model keywords to quickly locate the call frequency of a specific model.
Group	Filter by the Channel Group to which the request was actually routed.
Log Type	Includes all types, consumption, top-up, error, refund, etc.
Token Name	Filter by API key name to accurately calculate consumption across different development environments.
Request ID	Enter the unique Request ID generated by the platform to locate a single request.
Upstream Request ID	Enter the original Request ID returned by the model provider.

Real-Time Metrics Dashboard

Real-time throughput and consumption metrics are provided at the top of the filter bar:

Usage: Total Quota consumption statistics under the current filter criteria;
RPM (Requests Per Minute): Concurrent request rate in the last 60 seconds;
TPM (Tokens Per Minute): Token throughput rate in the last 60 seconds (Input Tokens + Output Tokens).

List and Details

The log list intuitively displays call time, model, consumed Quota, prompt tokens, completion tokens, and response latency. Click on an individual log entry to expand and view request headers, detailed error information, and context snapshots.

Drawing Logs

Navigate to /usage-logs/drawing from the left sidebar to view details of image generation tasks such as Midjourney.

Displays Prompt, submission time, duration, execution progress, and status;
Once generation is complete, images can be previewed directly in the list, with one-click support for subsequent actions such as Upscale and Variation.

Async Task Logs

Navigate to /usage-logs/task from the left sidebar to centrally track long-running asynchronous tasks such as video generation (Sora / Kling / Jimeng) and Suno music.

Task Status Description

Status	Detailed Meaning
Queued / In Queue	The task has been successfully submitted to the generation queue and is waiting for compute scheduling.
In Progress / Executing	The upstream generation engine is rendering/generating, and the progress percentage is queryable.
Success	The task is completed; you can preview it in the console or obtain the download URL via API.
Failed	Generation failed; the details column provides the specific error reason returned upstream.

Async task statuses are typically polled and updated by the background system every 15 seconds. Please wait patiently for status updates as large model video/audio rendering takes time.

Failures, Timeouts, and Automatic Refunds

If an asynchronous task ultimately fails due to upstream service disruptions or timeouts:

If the task pre-deducted Quota during the submission stage, the system will automatically refund the pre-deducted quota in full upon confirming the failure;
When the refund succeeds, the system will automatically record a "Refund" entry in the general usage logs and return the Quota to the original account balance or subscription quota.

API Reference

Read-Only Log Query APIs

In addition to the console, developers can directly fetch call logs via APIs for automated reconciliation:

Data Type	User Read-Only Endpoint	Time Parameter Unit
General Log Details	`GET /api/log/self`	Seconds (`created_at`)
Log Statistics Summary	`GET /api/log/self/stat`	Seconds
Midjourney Drawing Logs	`GET /api/mj/self`	Milliseconds (`submit_time`)
Async Task Logs	`GET /api/task/self`	Seconds (`submit_time`)

The endpoints support standard pagination parameters p (page number, starting from 1) and page_size (items per page, maximum 100).

Async Generation and Task Endpoints

Interface Type	Path	Description
Video Generation	`POST /v1/videos`	Submit an asynchronous video generation task and return `task_id`.
Video Task Query	`GET /v1/videos/{task_id}`	Query generation progress and result URLs.
Video Stream Retrieval	`GET /v1/videos/{task_id}/content`	Stream proxy to retrieve generated video binary content.
Kling Text-to-Video	`POST /kling/v1/videos/text2video`	Submit a text-to-video task via Kling protocol.
Kling Image-to-Video	`POST /kling/v1/videos/image2video`	Submit an image-to-video task via Kling protocol.
Suno Music Task	`POST /suno/submit/{action}`	Submit a Suno music creation task.
Suno Status Query	`GET /suno/fetch/{id}`	Get the playback URL and lyrics of the generated audio.
Midjourney Task	`POST /mj/submit/{action}`	Submit Imagine, Blend, Describe, and other tasks.
Midjourney Query	`GET /mj/task/{id}/fetch`	Query Midjourney task status and image results.

Relationship with Data Dashboard

Logs: Itemized raw request streams containing millisecond-level latency, errors, and precise Tokens for each call.
Data Dashboard: Macro trend charts aggregated by time period (day/week/month), model type, and API key.
Recommendation: When encountering single-call anomalies or billing questions, prioritize using Usage Logs to locate and troubleshoot via Request ID; refer to the Data Dashboard when evaluating overall usage and model selection costs.

📜 Logs and Async Tasks

On this page