Gemini 2.5 Flash Native Audio Preview 12-2025
Neural Network
Specifications
- Input
- text, images, documents
- Output
- text, audio
- Reasoning effort
- low, medium, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
Image prompt
(per 1K tokens)
How it works Gemini 2.5 Flash Native Audio Preview 12-2025?
Frequently asked questions about Gemini 2.5 Flash Native Audio Preview 12-2025
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
This is a Google Gemini model with native audio generation: it accepts text, images, and documents, and responds with text or audio. It is suitable for voice scenarios, voiceovers, analyzing files and screenshots, as well as tasks requiring reasoning and external tool calls. Context is 200,000 tokens.
In addition to text, the model outputs audio — it synthesizes speech based on your text description. The response limit is 32,000 tokens, so voluminous material fits in a single request. Specific parameters and output formats depend on how you access the model via the BotHub API.
Yes, reasoning before answering is confirmed for this model. It processes the task first and then provides the result — this is noticeable in multi-step questions, document analysis, and logical chains. In BotHub, you can immediately compare its answer with another model in the same window.
Yes, both function calling and structured JSON output are supported. The model decides when to call an external tool and returns the answer according to the specified schema. With the unified OpenAI-compatible BotHub API, you only need to write the integration once, and then you can switch models.
The model accepts images and documents as input: send a screenshot, photo, or file and ask it to analyze the content. Audio and video input is not noted in our data, so it is easiest to test this with your own material directly in the BotHub interface.
The Google Gemini family is known for broad language support, but we do not have data on prompt and output language specifically for this version, so we cannot make any promises: send a typical request and evaluate it yourself. The BotHub service itself is Russian-speaking, accepts Russian cards, and is accessible from the Russian Federation without a VPN.