The team behind the AI project Suno has introduced the v5 model for music generation. The update is available in beta for Pro and Premier subscribers.
Everything changes with Suno v5. Launching today for Pro and Premier subscribers, the world’s best music model delivers more immersive audio, authentic vocals, and unparalleled creative control that will transform how you make music.
This breakthrough goes beyond making better… pic.twitter.com/QNrci69JW2
— Suno (@SunoMusic) September 23, 2025
In the official announcement, developers stated that v5 offers “more immersive audio, authentic vocals, and unprecedented creative control.” They claim it is the third consecutive model to surpass external developments and is “the best to date.”
“With the release of Suno v5, everything changes. This breakthrough is not just about making better songs. v5 is the intelligent foundation for our most advanced tools, which will transform the music creation process,” the statement reads.
The new version also improves the recognition and blending of musical genres. Users have gained more options for customizing compositions.
Early listeners have already published tracks created with Suno v5 and have praised the quality of the generation.
Absolutely Incredible! What a huge upgrade!
I was saving these lyrics for awhile and it one-shot this! https://t.co/P7p7GoSZug
— Adam The AI Guy ✧ (@And2TheRepublik) September 24, 2025
The new model can be tried on the Suno website. The monthly subscription costs $10.
Vocals Become More Expressive, Guitars Still “Dirty”
Vsévolod, the lead singer of the band “Singular Decomposition,” shared his thoughts on Suno v5 with ForkLog. He noted significant improvements in vocal quality and ease of use but pointed out persistent issues with instruments and genre blending.
He identified the main difference between the new version and previous ones as the expression of vocalists and the correct placement of stress in words. According to him, the difference in vocal quality between v5 and v3 is “a colossal gap.”
“In the fifth version, you no longer need to write a word as it is spelled for Suno to sing it correctly. It is visually more pleasing to me when the word is immediately written correctly in the text,” explained Vsévolod.
Another important improvement is the handling of metatags. They can now be attached to an entire part of a song rather than placed before each line, saving time.
Progress in the instrumental part is less noticeable. According to the musician, the quality of electronic guitars in Suno v5 has become “slightly cleaner,” but they still sound “dirty and blend together.” Acoustic guitars, ambient components, and drum parts have not raised concerns since the fourth version. The model’s strong suit remains electronic music, as it is its “native element.”
Vsévolod noted that he anticipates the launch of the Studio feature for full track rearrangement within the service.
The musician believes that full style blending in Suno is not yet achievable. He explained that the result may be satisfactory when listening with simple headphones, but on high-quality equipment, the shortcomings become apparent.
“If you listen to music on AirPods, you’ll be satisfied after one listen. If you listen on Marshalls, it will start to hurt your ears,” he added.
The model performs well with genres that are close in spirit, such as nu-metal and alternative rock. However, attempting to combine opposing styles, like “Italian opera and Californian trap,” will result in “dirty” sound, as the neural network does not yet understand this, concluded Vsévolod.
Back in August, Andrey, the author of the YouTube channel “Oh, Guitarist!” discussed with ForkLog the art of neuro-covers and the significance of cryptocurrencies for the AI music market.