google

Google: Lyria 3 Pro Preview

Google: Lyria 3 Pro Preview is an audio-generating model from google, available at $0 per million tokens for both input and output during its preview period. It accepts text and image inputs and produces text and audio outputs, with a 1,048,576-token context window. As a specialized audio generation model, it falls outside the scope of our standard 12-benchmark LLM testing suite, which evaluates text-based capabilities. Accordingly, we do not have benchmark scores or an overall ranking for this model. Teams looking for a capable audio generation option from google can explore Lyria 3 Pro Preview at zero cost during the preview phase.

Performance

Google: Lyria 3 Pro Preview has not been included in our standard 12-benchmark testing suite. Our benchmarks evaluate text-based LLM capabilities such as classification, agentic planning, faithfulness, and constrained rewriting — categories that do not apply to audio generation models. We cannot report scores, rankings, or strengths and weaknesses for this model based on our testing. Teams evaluating Lyria 3 Pro Preview for audio generation use cases should conduct their own quality assessments on representative samples.

Pricing

Lyria 3 Pro Preview is currently priced at $0 per million tokens — making it free to access during the preview period. As a preview-stage model, pricing and availability terms may change when the model moves to general availability. For teams evaluating audio generation capabilities from google, the zero-cost preview window is an opportunity to assess quality before pricing is established. No cost estimates at different usage volumes are available because the current rate is $0.

google

Google: Lyria 3 Pro Preview

Overall
0.00/5N/A

External Benchmarks

SWE-bench Verified
N/A
MATH Level 5
N/A
AIME 2025
N/A

Pricing

Input

$0.00/MTok

Output

$0.00/MTok

Context Window1049K

modelpicker.net

Real-World Costs

iChat response$0.00
iBlog post$0.00
iDocument batch$0.00
iPipeline run$0.00

Pricing vs Performance

Output cost per million tokens (log scale) vs average score across our 12 internal benchmarks

This modelOther models

Try It

from openai import OpenAI

client = OpenAI(
    base_url="https://openrouter.ai/api/v1",
    api_key="YOUR_OPENROUTER_KEY",
)

response = client.chat.completions.create(
    model="google/lyria-3-pro-preview",
    messages=[
        {"role": "user", "content": "Hello, Google: Lyria 3 Pro Preview!"}
    ],
)

print(response.choices[0].message.content)

Recommendation

Google: Lyria 3 Pro Preview is relevant for teams building audio generation pipelines — music creation, sound effects, or other audio output use cases — who want access to google's audio model at no cost during the preview period. Since our benchmarks focus on text-based LLM performance, this profile does not capture the capabilities most relevant to audio generation. Teams with audio generation needs should evaluate the model directly. If you need a text-based model from google rather than an audio model, the Gemma 4 31B (avg 4.416 in our testing) or Gemini 3.1 Pro Preview (avg 4.333) are strong options in the google family.

How We Test

We test every model against our 12-benchmark suite covering tool calling, agentic planning, creative problem solving, safety calibration, and more. Each test is scored 1–5 by an LLM judge. Read our full methodology.