Qwen3-30B-A3B-Instruct-2507

Qwen3-30B-A3B-Instruct-2507 is an interactive language model specifically optimized for dialogue and instruction-following tasks. This updated version of Qwen3-30B-A3B employs the same Mixture-of-Experts (MoE) architecture with 30.5 billion total parameters and 3.3 billion active parameters, but omits explicit reasoning steps and is tuned for instantaneous response generation. Architectural enhancements include native support for an extended context length of up to 262,144 tokens, with a recommended output length of 16,384 tokens per generation.

The model achieves outstanding performance of 90.0 on the ZebraLogic benchmark, significantly surpassing both DeepSeek-V3 (83.4) and GPT-4o (52.6). This demonstrates the model’s strong ability to produce logically coherent and well-justified responses without relying on step-by-step reasoning. In creative tasks, the model excels with scores of 86.0 on Creative Writing v3 and 85.5 on WritingBench, outperforming GPT-4o and Gemini-2.5-Flash respectively, making it an unmatched tool for generating high-quality creative content. Additionally, the model performs exceptionally well in programming tasks, achieving 83.8 on MultiPL-E, showcasing its capability to generate high-quality code across various programming languages.

These features and capabilities make Qwen3-30B-A3B-Instruct-2507 the ideal choice for applications requiring lightning-fast responses and natural user interaction. It is perfectly suited for chatbots and applications focused on rapidly generating high-quality textual content.

Announce Date: 29.07.2025
Parameters: 31B
Experts: 128
Activated at inference: 4B
Context: 263K
Layers: 48
Attention Type: Full or Sliding Window Attention
Developer: Qwen
Transformers Version: 4.51.0
vLLM Version: 4.51.0
License: Apache 2.0

Use our pre-built public endpoints for free to test inference and explore Qwen3-30B-A3B-Instruct-2507 capabilities. You can obtain an API access token on the token management page after registration and verification.

An endpoint is a server with pre-installed software that automatically downloads neural network weights and launches an OpenAI-compatible API to handle incoming requests.

A private endpoint is created within the user's project, belongs exclusively to the user, and is billed per server configuration in accordance with the infrastructure pricing.

We recommend creating private endpoints when you need to:

• achieve maximum endpoint performance,
• access the full context window for processing large requests and conversations,
• ensure maximum security by processing data in a dedicated, isolated environment.

Contact our dedicated neural networks support team at support@immers.cloud or send your request to the sales department at sale@immers.cloud.