Gemini 2.5 Pro Preview
Neural Network
Gemini 2.5 Pro Preview is a Google model with step-by-step reasoning, suitable for code, math, and scientific tasks.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 1 048 576 tokens
- Max answer length
- 65 536 tokens
- 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 Pro Preview?
Gemini 2.5 Pro Preview is a Google model for advanced reasoning, programming, math, and scientific tasks. Its distinctive feature is the reasoning mode: before answering, the model works through the solution process, which significantly improves accuracy on multi-step tasks. Via BotHub, it is accessible from Russia without a VPN or foreign card; payment is made with Russian cards in rubles, based on actual token usage rather than a subscription, and tokens do not expire. Over 250 neural networks are gathered in one window, allowing you to compare responses from different models on a single query. A unified OpenAI-compatible API allows you to switch models in your service without rewriting the integration; dialogues are encrypted and not saved. Companies have access to contracts, invoices, electronic document management (EDO), and an admin panel with employee limits. The model is useful when you need to understand someone else's code and suggest refactoring, debug an obscure error, solve a complex math or engineering problem, analyze a scientific paper, or design a service architecture before development begins.Frequently asked questions about Gemini 2.5 Pro 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.
This is a Google text model with a context of 1,048,576 tokens: it analyzes large documents and codebases, writes and refactors code, finds errors, solves math and logic problems, answers based on image content, and connects to external tools via function calling.
The output limit is 65,536 tokens per response. This is enough to get a voluminous document, a detailed analysis, or a large code file in its entirety, without manually splitting the task. If the response hits the limit, ask the model to continue from where it stopped.
Yes, this is a confirmed capability: before answering, the model builds an internal chain of reasoning and only then provides the result. The difference is noticeable in multi-step tasks — logic analysis, math, finding the cause of a code error, planning complex architecture.
Yes, the model supports function calling and structured JSON output, making it easy to integrate into agents, chatbots, and pipelines. In BotHub, it is available via a unified OpenAI-compatible API: you can switch models without rewriting the integration.
Images and documents — yes, this is confirmed by our data: you can send a screenshot, diagram, PDF, or contract and ask for an analysis. There is no note in the data regarding audio and video, so we do not promise it in advance — check the available formats directly in the interface.
We do not have data on prompt and response languages, so we will not make any guarantees. It is easier to check with your own task: in BotHub, payment is based on actual token usage, no subscription is needed, and other models are available in the same window for comparison.