O3 Mini Batch
Neural Network
O3 Mini for code and STEM tasks — OpenAI reasoning model with a customizable reasoning_effort parameter.
Specifications
- Input
- text, documents
- Output
- text
- Context size
- 200 000 tokens
- Max answer length
- 100 000 tokens
- Reasoning effort
- low, medium, high
- Capabilities
- function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works O3 Mini Batch?
O3 Mini is a reasoning model from OpenAI designed for STEM: science, math, and programming. Its key feature is the reasoning_effort parameter, which lets you set the depth of reasoning: less for faster, cheaper responses; more for precise analysis of complex tasks. The batch variant is designed for batch processing of requests where volume matters more than instant response. In BotHub, the model is accessible from Russia without a VPN or foreign card, and you pay per token in rubles without a subscription. Over 250 models are available in the same window, making it easy to compare and switch, while a unified OpenAI-compatible API allows you to change models in your code with one line without rewriting the integration. Chats are encrypted via AES-GCM, and companies have access to contracts, invoices, EDI, and an admin panel with limits. It is useful when solving engineering or olympiad problems step-by-step, writing and refactoring code, debugging algorithm logic, running batches of similar calculations, or testing research hypotheses.Frequently asked questions about O3 Mini Batch
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 an OpenAI text model with a 200,000-token context window: document analysis and comparison, coding, analytics, STEM tasks, and summary preparation. The batch variant is convenient when you need to run a large volume of similar requests in a batch rather than one by one.
The output limit is up to 100,000 tokens per response. This is enough for extensive technical analysis, detailed documentation, large code snippets with comments, or a detailed summary of an uploaded document without breaking the task into parts.
The model belongs to OpenAI's o-series and supports the effort parameter: you decide how many resources to spend on thinking. For simple requests, minimal effort is more efficient; for math, logic, and code analysis, increased effort is better.
Yes, according to our data, function calling and structured JSON output are available. This allows integrating the model into agents and pipelines where the response needs to be passed directly to code, a database, or an external service without manual text parsing.
Besides text, the model accepts documents as input — they can be analyzed, compared, and summarized. For images, audio, and video, BotHub has separate models, and you can switch to them in the same window or via a unified API.
There is no specific note about languages in our data, so it is best to test the model with your own prompts and documents — in BotHub, this takes just a couple of clicks. The service itself works from Russia without a VPN, with payment via Russian cards in rubles.