Google DeepMind
Google DeepMind
Audio model · United States ·Jun 2026

Magenta RealTime 2

Audio Open Weights
2.4B
Parameters

About Magenta RealTime 2

Magenta RealTime 2 is an AI model developed by Google DeepMind, in the audio category, released in 2026, made available as an open-weights model with 2.4B params. An open-weights live music model that generates 48 kHz stereo musical audio in real time, responding to MIDI, text, and .

On this page you'll find Magenta RealTime 2's full specifications. Review provider pricing and benchmark scores below, or compare Magenta RealTime 2 head-to-head with other audio models.

→ More models by Google DeepMind → All audio models → AI benchmarks leaderboard

Run Magenta RealTime 2 locally — what fits

GPUs with enough VRAM to hold Magenta RealTime 2 at FP16 (lower precision needs less), from best value to most headroom. Generation speed depends on your tool, resolution and step count, so we don't estimate it.

Good
Intel Arc B580 12GB
Consumer GPU · ~$249
12 GB VRAM
fits at FP16 · ~4.7 GB headroom
See fit detail →
Better
AMD Ryzen AI Max+ 395
Consumer GPU · ~$1,999
128 GB VRAM
fits at FP16 · ~108.9 GB headroom
See fit detail →
Best
Apple M2 Ultra 192GB
Apple · ~$6,600
192 GB VRAM
fits at FP16 · ~166.5 GB headroom
See fit detail →

VRAM fit only — we don't estimate generation speed for image/video/audio models (it depends on the tool, resolution and steps).

Consumer GPU
32 options that fit
→
Apple
18 options that fit
→
Datacenter GPU
13 options that fit
→

Community

Would you run Magenta RealTime 2 again?
No votes yet — be the first.
Did it run? — community reports by hardware

No reports yet. Own this model on some hardware? Be the first to confirm it runs.

Discussion

No comments yet. Start the discussion.

More models like Magenta RealTime 2

Google DeepMind
DiffusionGemma 26B-A4B
Google DeepMind · 26B
 Google DeepMind
Gemini 3.5 Live Translate
Google DeepMind
 Google DeepMind
Gemma 4 12B
Google DeepMind · 12B
 Google DeepMind
Gemini Omni
Google DeepMind
 Google DeepMind
Gemini 3.5 Flash
Google DeepMind
 Google DeepMind
Gemma 4 31B
Google DeepMind · 31B

Frequently asked questions

What is Magenta RealTime 2? ▶

An open-weights live music model that generates 48 kHz stereo musical audio in real time, responding to MIDI, text, and It is tracked on GenAIList with its specifications, benchmark scores and provider pricing.

Who created Magenta RealTime 2? ▶

Magenta RealTime 2 was developed by Google DeepMind and released in 2026.

Is Magenta RealTime 2 open source or proprietary? ▶

Magenta RealTime 2 ships as an open-weights model: you can download and self-host the weights, though the licence may place some restrictions on use.

How much does Magenta RealTime 2 cost? ▶

Pricing for Magenta RealTime 2 depends on the provider. See the providers table on this page for the latest API rates.

How does Magenta RealTime 2 perform on benchmarks? ▶

Benchmark scores for Magenta RealTime 2 are listed on this page as they are published. You can compare it head-to-head with other models on the GenAIList compare tool.

Google's New AI Music Model is Insane! (Magenta RealTime 2)
PREVIEW: magenta real-time 2 with DEMON backend

Know a great video about Magenta RealTime 2?