Alibaba: qwen3-30b-a3b-instruct-2507 API
A flagship Mixture-of-Experts (MoE) instruction-tuned snapshot of the Qwen3 series, released by Alibaba Cloud on July 30, 2025. Featuring 30.5B total parameters with only 3.3B activated per token (A3B), this 2507 Instruct version is optimized for Non-Thinking tasks. By internalizing reasoning logic, it delivers near-instant responses while maintaining performance parity with GPT-4o on rigorous benchmarks like MMLU-Pro and GPQA. It marks a significant leap in multilingual knowledge, instruction following, and long-context stability, making it the premier choice for high-performance RAG systems, real-time chatbots, and sophisticated AI Agents.
- Context window: 262,144 tokens
- Max output: 16,384 tokens
- Input: text
- Output: text
- Reasoning: Supported
- Tool calling: Supported
- Released: 2025-07-30
- Knowledge cutoff: 2025-05-31
Frequently Asked Questions
What is the context window of qwen3-30b-a3b-instruct-2507?
qwen3-30b-a3b-instruct-2507 supports a context window of up to 262,144 tokens.
Does qwen3-30b-a3b-instruct-2507 support function calling?
Yes. qwen3-30b-a3b-instruct-2507 supports tool / function calling.
Does qwen3-30b-a3b-instruct-2507 support reasoning?
Yes. qwen3-30b-a3b-instruct-2507 is a reasoning-capable model.
What is the knowledge cutoff of qwen3-30b-a3b-instruct-2507?
The knowledge cutoff of qwen3-30b-a3b-instruct-2507 is 2025-05-31.