Best Audio Generation Models
Model rankings updated based on real usage data.
Models are ranked by total prompt and completion tokens processed through the OpenRouter API over the trailing 7 days. Showing the top 4 models. Rankings measure usage on OpenRouter and reflect adoption, not model quality or benchmark performance.
Audio generation models create audio output from text or other prompts, powering use cases like music generation, sound design, voice-enabled assistants, and multimodal applications that respond with audio. This collection highlights some of the best audio generation models available on OpenRouter, making it easier to compare quality, pricing, and latency across providers through a single API.
Top Audio Generation Models on OpenRouter
OpenAI: GPT Audio Mini
465M tokens
A cost-efficient version of GPT Audio. The new snapshot features an upgraded decoder for more natural sounding voices and maintains better voice consistency. Input is priced at $0.60 per million tokens and output is priced at $2.40 per million tokens.
Google: Lyria 3 Pro Preview
207M tokens
Full-length songs are priced at $0.08 per song. Lyria 3 is Google's family of music generation models, available through the Gemini API. With Lyria 3, you can generate high-quality, 48kHz stereo audio from text prompts or from images. These models deliver structural coherence, including vocals, timed lyrics, and full instrumental arrangements. Lyria 3 Pro can generate full-length songs with verses, choruses, bridges.
OpenAI: GPT Audio
104M tokens
The gpt-audio model is OpenAI's first generally available audio model. The new snapshot features an upgraded decoder for more natural sounding voices and maintains better voice consistency. Audio is priced at $32 per million input tokens and $64 per million output tokens.
Google: Lyria 3 Clip Preview
9.05M tokens
30 second duration clips are priced at $0.04 per clip. Lyria 3 is Google's family of music generation models, available through the Gemini API. With Lyria 3, you can generate high-quality, 48kHz stereo audio from text prompts or from images. These models deliver structural coherence, including vocals, timed lyrics, and full instrumental arrangements. Lyria 3 Clip can generate short clips, loops, previews.
Frequently asked questions
Send POST /api/v1/chat/completions with modalities set to ["text", "audio"] and an audio object naming the voice and format. Audio output requires stream: true; the response arrives as SSE chunks, and the client concatenates the base64 delta.audio.data fragments in order before decoding the combined string once into the final file. The audio guide covers the request and the chunk format. For dedicated speech synthesis, use the text-to-speech models at POST /api/v1/audio/speech.
Listings with published pricing show input audio and output audio prices per million tokens next to their text token prices, so the cost of a reply depends on how much audio it returns. You pay only for the requests you make, with no separate subscription per provider.
Set audio.voice and audio.format in the request. Voices such as alloy or nova and formats such as wav, mp3, flac, opus, and pcm16 vary by model, so check the model page before you choose.
Yes. One OpenRouter API key works for every audio model on this page and for every other model on OpenRouter. Create a key on your API keys page.