The best AI music and audio models in 2026

The best AI music and audio models right now are ACE-Step 1.5, Eleven Music v2.5 and GPT-Live-1 (and GPT-Realtime-2.1), out of 8 we have reviewed. Each is ranked by our rating, with its price, our verdict and who it suits.

By Andrew Miller · Updated Nov 7, 2026 · How we review

How we ranked: by our rating out of 5, from reviews built on each product's own pricing, documentation and policies plus independent benchmarks where they exist, each checked on the date shown in the review. We take no affiliate commission.

1. 1
   A

   ACE-Step 1.5

   ACE-Step 1.5 is the open music model to start with: MIT-licensed, fast on a modest GPU and full of editing tools, even if the newest open rivals sound better.

   Best for: Developers, game studios and video makers who need commercially usable songs from a model that runs on an ordinary GPU or a Mac.

   Price: Free under the MIT licence, commercial use included. The team's hosted app at acemusic.ai is free. The turbo model self-hosts on a GPU with under 4 GB of VRAM.

   Read our ACE-Step 1.5 review
   4.3/5
2. 2
   E

   Eleven Music v2.5

   Eleven Music v2.5 costs more per song than Lyria, but it lets you rewrite a single section and comes with the clearest commercial terms of any music API we cover.

   Best for: Marketing teams and app developers who need publishable songs with vocals through an API and want to edit sections rather than regenerate whole tracks.

   Price: API: $0.15 per minute of music on every plan; Finetunes $1.50 each. Included music minutes: Free 3, Starter ($6/month) 40, Creator ($22/month) 147, Pro ($99/month) 660, Scale ($299/month) 1,993, Business ($990/month) 6,600. Commercial use from Starter.

   Read our Eleven Music v2.5 review
   4.3/5
3. 3
   

   GPT-Live-1 (and GPT-Realtime-2.1)

   OpenAI's most natural-sounding way to build a phone agent, with interruptions handled properly; the bill is simpler per minute but not necessarily smaller.

   Best for: Teams building high-volume phone support or sales agents who want natural turn-taking and freedom to choose the reasoning model.

   Price: GPT-Live-1 $0.05 per session minute, billed per second, plus the backend model's tokens. GPT-Realtime-2.1 audio $32 input / $64 output per 1M tokens ($0.40 cached); mini $10 / $20. GPT-Realtime-Translate $0.034 a minute.

   Read our GPT-Live-1 (and GPT-Realtime-2.1) review
   4.3/5
4. 4
   G

   Grok Voice 3.0

   Grok Voice 3.0 leads the independent speech-to-speech index at a simple flat rate, but OpenAI still wins on support-style tasks and every call is processed in a single US region.

   Best for: Teams building phone and app voice agents who want the strongest benchmark scores at a predictable per-minute price and can work with a US-only deployment.

   Price: $0.08 per minute of audio ($4.80 an hour) plus $0.004 per text input message, the same as Grok Voice 2.0, with reasoning included in the per-minute rate.

   Read our Grok Voice 3.0 review
   4.3/5
5. 5
   L

   Lyria 3.5

   Lyria 3.5 is the music model to call from code if you need full songs with vocals at a flat price, though it gives you one take per prompt and no way to edit it.

   Best for: Developers who need full-length songs with vocals generated inside their own app at a predictable per-track price.

   Price: Gemini API paid tier only: Lyria 3.5 $0.08 per song; legacy Lyria 3 Pro Preview $0.08 per song and Lyria 3 Clip Preview $0.04 per 30-second clip. No free API tier. Also in the Gemini app and Flow Music on Google's consumer plans.

   Read our Lyria 3.5 review
   4.2/5
6. 6
   M

   MiniMax Music 3

   MiniMax Music 3 sings your lyrics clearly for up to five minutes and lets you sell the result, provided you have a 24 GB NVIDIA card and put its name on your product.

   Best for: Companies that want to self-host a commercial song generator with clear vocals and can provide a 24 GB NVIDIA GPU.

   Price: Weights free under the MiniMax-Music3 Community License, commercial use included; written authorisation needed above $20 million annual revenue. MiniMax API: $0.15 per track up to five minutes, existing paying users only since 20 August 2026.

   Read our MiniMax Music 3 review
   4.2/5
7. 7
   Y

   YuE2-3B

   YuE2-3B is the open song model to study and tinker with, because you can edit its score note by note, but its non-commercial licence keeps it out of anything you sell.

   Best for: Researchers, hobbyists and songwriters working non-commercially who want to revise a song's melody and chords rather than regenerate it.

   Price: Free weights under CC BY-NC 4.0: non-commercial use only. No hosted API; you pay for your own GPU.

   Read our YuE2-3B review
   4.0/5
8. 8
   S

   Stable Audio 3.0 Medium

   Stable Audio 3.0 Medium is the open choice for instrumentals and sound effects trained on licensed audio, with small siblings that run on a phone, but it does not make songs with lyrics.

   Best for: Game, app and video teams that need instrumental music or sound effects from a model with documented, licensed training data, including on device.

   Price: Weights free under the Stability AI Community License for organisations under $1 million annual revenue; above that, an Enterprise licence at custom pricing. API: Stable Audio 3.0 costs 8 credits ($0.08) per generation; new accounts get 25 free credits.

   Read our Stable Audio 3.0 Medium review
   3.9/5