Qwen3 30B A3B
Neural Network
Qwen3 30B A3B is a Qwen model based on MoE architecture for reasoning, multilingual tasks, and agent scenarios.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 40 960 tokens
- Max answer length
- 16 384 tokens
- Reasoning effort
- low, medium, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Qwen3 30B A3B?
Qwen3 30B A3B belongs to the Qwen3 generation and is built on the Mixture of Experts (MoE) architecture, where only a portion of parameters is used for each request. The Qwen team focuses on reasoning, multilingual support, and advanced agent tasks in this line, making it ideal for scenarios requiring step-by-step chains rather than short answers. In BotHub, Qwen3 30B A3B is accessible from Russia without a VPN or foreign card: pay with a Russian card in rubles only for used tokens, with no subscription, and unused balance does not expire. Over 250 models are available in one window for easy comparison, and a unified OpenAI-compatible API allows you to swap models in a multi-model system without rewriting integration. Chats are encrypted, and companies have access to contracts, invoices, electronic document management, and an admin panel with limits. It is useful for multi-step reasoning tasks, coding and debugging, building agents with function calling, working with documents and correspondence in different languages, or running batch requests via API in your service.Frequently asked questions about Qwen3 30B A3B
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.
qwen3-30b-a3b works with text and documents: analysis and editing, coding and debugging, analytics, and answers based on uploaded files. The model can reason before answering and call external functions, making it suitable for both agent scenarios and daily work tasks.
The model generates up to 16,384 tokens per response — enough for a long article, detailed analysis, or a large code module. The total context window is 40,960 tokens, including your request with files and the response itself. If you need longer text, break the task into parts.
Yes, reasoning before answering is confirmed by our data. The model first breaks down the condition step-by-step and then provides a result, which significantly helps with math, logic, and code debugging. For simple questions, it barely affects usability, but for complex ones, it increases accuracy.
Yes, the model supports function calling and structured JSON output. You can connect it to your tools, databases, and services via the unified OpenAI-compatible BotHub API, and switch to another model if needed without rewriting the integration.
The model accepts images and documents as input — you can ask it to describe an image, extract data from a scan, or analyze a contract. Audio and video are not supported in our data, so for transcription or audio processing, it is better to choose a specialized model from the BotHub catalog.
Qwen3 is a multilingual family, and in practice, the model works confidently with Russian-language tasks. The exact result depends on your scenario, so it is easier to check with your own examples: in BotHub, you can open it in one window alongside other models and compare answers without a VPN or foreign card.