gemini-robotics-er-2-preview
Neural Network
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 131 072 tokens
- Max answer length
- 32 000 tokens
- Capabilities
- Batch API
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
Image prompt
(per 1K tokens)
How it works gemini-robotics-er-2-preview?
Frequently asked questions about gemini-robotics-er-2-preview
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.
The Google model accepts text, images, and documents as input and responds with text. The context window is 65,536 tokens, so a large document or a series of frames will fit into a single request. It is suitable for image analysis, data extraction from files, and text analytics based on visual material.
The limit for a single response is 32,000 tokens, which is approximately several dozen pages. This is more than enough for most tasks: detailed document analysis, long summaries, or structured descriptions of a series of images. If you need a larger volume, break the work into sequential requests — the 65,536-token context is sufficient for this.
There is no specific note about a reasoning mode in our data, so we will not promise it. However, nothing stops you from asking the model to break down the task step-by-step directly in the prompt — this usually results in more accurate answers. The easiest way to check is with your own example.
There is no note about function calling or strict JSON mode in our data, so we will not guarantee them. You can request the desired format with a standard instruction in the prompt and a schema description. The model is available via the unified OpenAI-compatible BotHub API, so testing your scenario is a matter of a couple of minutes.
Images and documents — yes, these are confirmed input formats: you can send a frame, scan, or file and get a text analysis. Audio and video are not listed as supported inputs, so it is better to use a different model for sound and clips — switching in one BotHub window is quick.
We do not have data on prompt and response languages, so we will not make any claims about Russian — this can only be verified with your specific task. What is definitely available: interface and payment in rubles, access from Russia without a VPN or foreign card, and pay-per-token billing.