useapi.net
215 subscribers
27 photos
27 videos
1 file
98 links
Third-party API for Midjourney, Mureka (Suno/Udio competitor), Runway, MiniMax, PixVerse, InsightFaceSwap and Pika
Visit https://useapi.net
Discord https://discord.gg/w28uK3cnmF
Reddit https://reddit.com/r/aiAPI
Download Telegram
Media is too big
VIEW IN TELEGRAM
PixVerse API v2 has been updated to support PixVerse V4 (part II):
• Create video using first and last frames POST videos/create-frames-v4.
• Extend video POST videos/extend-v4.
• Restyle video POST videos/restyle-v4.
• Lip sync video POST videos/lipsync, updated to use V4.
EXAMPLES
Mureka API v1 has been updated to support the recently released model O1, enable stems download and generate instrumental music.
• Currently supported models: O1, V6 and V5.5.
• Create instrumental music POST music/create-instrumental.
• Download the song license and stems POST music/download.
MiniMax API v1 TTS (text-to-speech) has been updated to support the recently added speech-02-hd and speech-02-turbo models. The speech-02-hd model is now the new default for all TTS audio/ endpoints.
PixVerse API v2 has been updated to support Off-Peak Mode, and a few small fixes have been applied to effects default sound generation.
Midjourney API users:
• When turning on the V7 model, you MUST complete the personalization training on their website. Until that step is completed, you will not be able to use V7. Once you have completed the training, you can go back and turn Personalization off via /settings.
• Another catch: as of now, V7 does not support standard mode, it only supports --turbo or --relax. If you have a standard account where --relax is not supported, all generations will be executed in --turbo mode even though the settings indicate that --fast mode is selected. This is likely to change soon and work as it did with V6.
New Subscription Model

We updated our subscription model, please take a look to familiarize yourself. Feel free to reach support if you have any questions or concerns.

1️⃣ Starting Thursday, April 10th, 2025, you will receive a 402 error with a detailed error message when attempting to add or update your AI API account if the number of your accounts exceeds your subscription quantity. Simply delete unused accounts to bring the total number below three, or update your subscription quantity.

2️⃣ Starting Thursday, April 17th, 2025, you will receive a 402 error with a detailed error message when using AI API endpoints if the number of your accounts exceeds your subscription quantity. Simply delete unused accounts to bring the total number below three, or update your subscription quantity.

👉 Check your useapi.net account status and current number of configured accounts to ensure uninterrupted service. We also sent an email with detailed instructions to all users affected by this change.

ANNOUNCEMENT
The following updates for Runway API v1 have been released:
• Upscale to 4K the Gen-4, Gen-4 Turbo, Gen-3 Alpha and Gen-3 Alpha Turbo videos POST gen4/upscale.
The following updates for Runway API v1 have been released:
• Gen-4 Turbo » Create a video using the first image(s) with a text prompt POST gen4turbo/create.
⚠️ Riffussion API has been decommissioned.
Currently we do not have plans to re-release it and recommend switching to Mureka API for music generation.
👉 PixVerse has updated its internal API for old v3.x routes - you'll get a 404 if you try to call an old 3.x route.
Consider switching to v4.
We will update the v3.x routes tomorrow to match the internal API changes.
PixVerse removed v3x support for effects and lipsync, and we will be dropping v3x support as well.
Here’s the PixVerse API v2 v3x to v4 migration map:
• POST videos/create-v3 → POST videos/create-v4.
• POST videos/create-frames-v3 → POST videos/create-frames-v4.
• POST videos/extend-v3 → POST videos/extend-v4.
🚀 We’re excited to announce the release of Kling API v1, our latest experimental API for Kling AI.
Kling offers text-to-video, image-to-video, and image manipulation capabilities with multiple model versions 1.5, 1.6 and 2.0
Key features include video generation from elements, special effects, virtual try-on, video extension, lip-syncing, and text-to-speech.
The API supports both free and paid Kling accounts.
As a bonus, you can execute an unlimited number of TTS generations for free via POST tts/create.
This feature is available for all Kling subscription plans, including the free one.
Kling API v1 has been updated to support KOLORS v1.5 and v2.0:
• Create images using KOLORS POST images/kolors
• Detect faces in images (free for all users) POST images/recognize-faces
• Upscale previously generated images POST images/upscale

EXAMPLES
Runway API v1 maintenance updates:
• Gen-2 endpoints are deprecated since they are no longer available on the Runway website.
• Get image description GET frames/describe response model updated.
⚠️ We suspect that Runway recently started targeting accounts with a large number of image/text-to-video Gen-3 Turbo generations. We're seeing some of these accounts get banned, while accounts with a comparable or greater number of Gen-4 generations are not affected. They probably figured that if you're still running a large number of Gen-3 Turbo generations, you're probably using automation, since humans would naturally switch to the more modern and efficient Gen-4.

We strongly suggest switching to Gen-4 for image/text-to-video going forward.