Suno 3.5 turns any sound into a four-minute song

Generating a complete song with AI used to feel like an exercise in frustration, largely because you ran out of time right as the groove got going.

Suno just rolled out version 3.5 of its AI music generator, and it takes aim directly at that limitation. The flagship upgrade is straightforward: you can now generate up to four minutes of audio in a single go.

Four-minute tracks in a single go

Previously, crafting a standard-length song meant stitching together shorter clips. With version 3.5, users can spit out full-length tracks with proper intros, verses, and hooks right out of the box.

If four minutes is not quite enough, the tool also allows extending songs by up to an additional two minutes. According to Suno, song structure has also received major improvements, helping prevent tracks from losing their musical coherence midway through.

Pro and Premier subscribers can access version 3.5 immediately, with free accounts getting access in a few days. Paying members can toggle between versions 3.5, 3, and 2, and export their audio in uncompressed WAV format. Suno noted that version 4 is already in the works.

From random sounds to full songs

Alongside the longer runtimes, Suno teased an upcoming feature that lets users pair any audio input with a text prompt to build a song.

Basically, you will be able to take a quick voice note, a rough rhythm, or an ambient sound, throw some style instructions at it, and let the model flesh out a full arrangement. Suno says this sound-to-song capability will be available soon.

Why it matters

For content creators, indie game developers, and bedroom producers, this turns AI music from an experimental toy into something genuinely functional.

Stitching together two-minute generations was clunky and often resulted in abrupt genre shifts. Moving to four-minute generations effectively solves the structure problem for standard pop, rock, and background tracks without endless rerolls.

We think Suno is making smart, product-first choices here. Offering version switching and WAV downloads shows respect for creators who actually need to drop these stems into a broader workflow. The real test will be whether the consistency holds across the entire four minutes without degrading into sonic mush, but the bar for AI music generation just got much higher.