Flixly dev login guide

Many users search for dev login thinking they need a separate portal, yet the actual path runs through the same account system that powers the main dashboard. After registration you gain immediate access to API keys that let you call any of the 50+ models without leaving the platform.

The direct route starts at the account settings once you sign up. From there you generate a personal token that works for all endpoints, including text-to-video calls that accept 1080p output at 8-second durations.

Account setup flow

Create the token right after the first credit purchase. Tokens last 90 days and can be rotated without losing queued jobs. Each generation still deducts credits at the published rate: 12 credits for a Seedance 2.0 720p clip, 18 credits for Veo 3.1 at 1080p.

Link the token inside your own code by sending it in the Authorization header. No extra dashboard pages exist for this step.

Text to Video calls accept the same token format as image endpoints. The same holds for Image to Video requests that reference an uploaded 512x512 PNG.

Integration examples

Developers commonly test the token first against the music endpoint before moving to longer video jobs. A minimal request body for a 4-second clip looks like this:

{
  "model": "kling-3.0",
  "prompt": "slow pan across neon city",
  "duration": 4,
  "resolution": "720p"
}

Credit usage appears in the response header so scripts can pause when the balance drops below 50 credits.

Tradeoffs to consider

The single-token system means one compromised key affects every tool. Rotate keys on a 30-day cycle if you expose them in CI pipelines. No granular scopes exist yet, so teams often create separate accounts for staging and production.

Voice Cloning and Lip Sync Video share the same auth layer, which keeps credential management simple but limits per-tool isolation.

Model availability after login

Once authenticated you can immediately call Frontier 2026 models without extra approval steps. Seedance 2.0 supports up to 10-second outputs. Wan 2.7 handles 24 fps at 1080p. Gemini 3.1 Flash TTS returns 30-second audio files in under 4 seconds.

Reference to Video jobs require the same token plus an image upload under 5 MB. Failures return standard HTTP codes so error handling stays straightforward.

Monitoring usage

The dashboard shows live credit burn per token. Export logs as CSV for 90-day windows. No separate billing portal is required; credits are deducted in real time.

Auto Captions and Shorts Generator both accept the dev token, letting one script chain captioning after video generation without extra logins.

The single decision rule is to rotate the token every 30 days and monitor the credit header on every response. That pattern keeps both security and spend under control while using the full set of models.

Environment Variable Management

Store the generated token in a dedicated variable rather than hard-coding it in scripts. Use a .env file loaded at runtime so the same codebase works across local machines and deployment targets without exposing the value in version control. Name the variable consistently, such as FLIXLY_TOKEN, and load it once at application start. Scripts that run in CI pipelines should pull the token from the platform’s secret store instead of committing it to the repository.

When rotating tokens every thirty days, update only the environment variable on each server or container. No code changes are required if the variable name stays the same. Keep a short-lived backup token in a separate variable during the rotation window so queued jobs continue without interruption.

Error Handling Patterns

Responses return standard HTTP status codes. A 401 indicates an expired or invalid token and should trigger an immediate retry with a fresh token pulled from the environment. A 429 signals that the credit balance has dropped below the threshold shown in the response header; pause the script and poll the balance endpoint until credits are replenished.

Wrap every call in a retry loop that backs off exponentially on transient network errors. Log the full response body on failures so the exact model parameter that caused the rejection can be identified later. Error Reference lists the numeric codes returned by each endpoint.

Workflow Automation Checklist

• Load token from environment on startup
• Check credit header before submitting batches larger than ten jobs
• Rotate token on the first day of each month using the account settings page
• Export the previous month’s CSV log before rotation so usage history is preserved
• Test a single music endpoint call after any token change to confirm authentication
• Chain captioning and video generation inside one script so only one token is needed for the full pipeline

Place the checklist in the project README so new team members follow the same sequence. Shorts Generator jobs can be added to the same script once the initial video call succeeds.

Production Deployment Notes

Run token rotation through a scheduled job that calls the account endpoint rather than manual dashboard visits. Store the new token in the secret manager first, then restart containers so the updated value is picked up. Keep the old token active for an additional twenty-four hours to cover any in-flight requests.

Separate staging and production accounts remain the simplest way to isolate spend when granular scopes are unavailable. Point the staging environment at its own token while production continues using the primary one. Both accounts share the same model list, so no additional approval steps are needed after the second account is created.

Track header values in application metrics so credit usage trends appear alongside latency and error rates. When the balance header falls below fifty credits, emit an alert before new jobs are queued. This pattern prevents unexpected pauses during overnight batch runs.

Automated Token Management

Writing a small rotation script removes the need for manual dashboard visits. A Python helper can fetch a fresh token via the account endpoint, store it in the secret manager, and restart the relevant containers. The script checks the response for the new expiration date before committing the change. Keep the rotation job on a separate schedule from production workloads so it runs during low-traffic windows.

Store both the primary token and a short-lived backup in the same secret store entry. The backup is activated only if the primary rotation fails or if queued jobs are still running. After a successful rotation, the script exports the prior month’s usage CSV and emails the link to the team alias for record keeping.

Test the rotation script in a staging account first. Point the staging jobs at their own token while the script updates the production secret. This separation prevents a failed rotation from halting live video generation pipelines.

Rate Limit and Credit Threshold Handling

Response headers return both remaining credits and the current rate-limit window. Scripts should read these values before submitting batches larger than five jobs. When credits fall below the 50-credit mark, the script pauses and polls the balance endpoint every thirty seconds until the threshold is restored.

Implement a sliding-window counter that tracks credits spent over the last sixty minutes. If the counter predicts depletion before the next scheduled top-up, the workflow can downgrade resolution or shorten duration automatically. This keeps overnight batches from stopping mid-run.

A simple table of common header values helps teams interpret responses quickly:

Log every header value alongside the model name and job duration. Over time the logs reveal usage patterns that inform better batch sizing.

CI/CD Integration Patterns

Store the token in the platform’s native secret store rather than repository variables. On each deployment the pipeline pulls the current value and injects it into the container environment. If the token has rotated since the last deploy, the pipeline detects the mismatch through a lightweight health-check call and fails the build with a clear message.

Add a post-deployment step that runs a single music-endpoint test. The test confirms authentication and prints the credit header so the deploy log shows the starting balance. Include the same test in the rollback playbook so reverting to an earlier image still uses a valid token.

Teams that run multiple regions keep one token per region. Each region’s secret is updated independently, allowing staged rollouts without global credential changes. Account Settings lists all active tokens so operators can revoke a single region without affecting others.

Logging and Audit Trail

Append the token’s last-four characters to every log line instead of the full value. This lets support staff correlate errors with specific keys without exposing secrets. Export the full CSV log weekly and store it alongside deployment records so usage can be audited against billing statements.

When an error occurs, capture the request body, response headers, and the model parameter. Store these details in a dedicated error table so patterns across jobs become visible. Error Reference provides the numeric codes that map to each failure type, making automated classification straightforward.

Rotate the audit log retention period independently of token rotation. Keep three months of detailed logs while discarding raw request bodies after thirty days to limit stored sensitive data.