MiniMax API v1 endpoint POST videos/create updated to support the End Frame parameter end_frame_fileID.

Example.

For the last four weeks, we have been notifying via email users who were still using the Midjourney API without setting up configuration accounts (aka Midjourney API v1, circa 2023).
According to our stats, there are a dozen active users who are still using the old way of calling the API and have ignored the warning emails.
We will be removing this ability starting September 1, 2025, so if you ignored those emails, you will be getting HTTP response 402:

{
  "error": "Configuration for channel <channel> not found. Please create a Midjourney API account at https://useapi.net/docs/api-v2/post-account-midjourney",
  "code": 402
}

You can follow the detailed instructions in the link provided in those emails.
In case you missed it: https://useapi.net/announcement-250808
It takes about 10–15 minutes to add configuration and update code so it no longer sends discord token with every call.
Please feel free to ask if you have any questions or concerns.

Midjourney API v2 has been updated:
• Upscaled image URLs are now available in the imageUx array. There is no longer a need to execute POST jobs/seed_async to retrieve upscaled images.
• When using --video upscaled video URLs are now available the videoUx array.
See both videoUx and imageUx in GET jobs/<jobid>.
• Prompt length for POST jobs/imagine has been extended from 2048 to 5000 characters.

⚠️ Security consideration when using videoUx and imageUx URLs:
URLs in both arrays follow this format: https://cdn.midjourney.com/<GUID>/…. They contain a job GUID which can be traced back to your account and will expose your account details such as your Discord account name and email. Exercise caution when posting these URLs publicly. You may want to save the URL content locally and serve it from your own origin, or proxy the links to obfuscate the original URLs.

Example.

This weekend a few users reported that they were getting 401 errors when calling the /kolors endpoints:

{
    "result": -401,
    "error_msg": "token value error",
    "error_id": "E1234567890"
}

That 401 is from the Kling internal API.
After re-adding the same account the error went away.
This is because when you add an account we always get a fresh token.
It appears that Kling expired a few session tokens ahead of schedule causing our API not to refresh them and resulting in the error above.
It's not clear at the moment if this is just an issue on Kling's side or if they are trying to terminate sessions based on certain activity.
Please let us know if this happens again so we can implement a workaround if this turns out to be a permanent issue.

Runway API v1 added POST gen4/act-two-voice endpoint for AI voice swapping with Act-Two.

Example.

We implemented Kling forced token expiration mitigation logic, which will handle the case above and will refresh the token for you.
We tested it extensively and hope this will be a solid solution going forward.
If you notice anything odd or out of order, please let us know.

Midjourney API v2 added GET proxy/cdn-midjourney endpoint to proxy asset retrieval from Midjourney CDN. Use this endpoint to download imageUx and videoUx URLs returned by GET jobs/jobid.

We have a working prototype of the Real-Time Midjourney API.
You read that right—it provides the same experience as your browser, allowing you to instantly see results without any delay.
We're hoping to release it in the very near future and are trying to collect feedback.

This will be an addition to our long-standing and very popular third-party Midjourney API v2.

The API returns an SSE text/event-stream stream with a live data feed, so we are not planning to release a callback feature, as it is somewhat pointless in this case.

We also want to ask how important the webhook/callback feature is for you, as we're not sure how well SSE streaming is supported by make.com or n8n—both have almost no documentation on that.
It is similar to MiniMax v1 LLM API and is commonly used by LLM APIs.

The API is extremely fast and will allow you to post multiple jobs to the same channel as fast as the Discord rate limit allows (usually 1–2 interaction events per second). This can be a big advantage for folks looking for ultimate performance, but also poses challenges, as this may be too fast and cause bans. Naturally, we can optionally allow you to apply desired throttling if you choose.

This API will also support uploading images directly to the Discord CDN, so you can host your images there and refer to those hosted URLs from your prompts.

It would be very helpful for us if you could answer the questions below:
1. I'm interested in the Real-Time Midjourney API and do not care about the webhook/callback feature at all.
2. I'm interested in the Real-Time Midjourney API, but the webhook/callback feature is a must for me.
3. I do not care for the Real-Time Midjourney API at all.

MiniMax released changes to their website. We're working on the fix, I'll keep you posted on our progress.

MiniMax files upload fix tested and deployed, please give it a try.

PixVerse recently added the ability to generate images rather than videos.

The GET videos/effects endpoint now returns two new fields:
• template_type: 1 | 2, 1 - video response, 2 - image response (new)
• template_model: `` | image_v6, image_v6 (new) for template_type: 2
There are only 3 image response templates for now: Sharkie Plush Delivery, Self-Figurine Roast and Your game of life, archived forever.

We added a quick fix to deal with the new template_type: 2. The POST videos/create endpoint, when an image template_id is used, will properly place the image generation task and respond with
"image_id": "user:12345-pixverse:user@email.com-image:123456789"
You can use the returned image_id value to retrieve the generated image via GET videos/<image_id> or delete it via DELETE videos/<image_id>.

We will monitor how PixVerse plans to support image generation—they will most likely add more soon, so we will move the above temporary image retrieval/delete endpoints to designated pixverse/images/ endpoints.

Mureka released a website update, attempting to generate V7 songs currently returns 400. We'll be releasing an update to fix this issue today. Meanwhile you can use V6.