Qwen3 235B A22B Thinking

System Requirements

VRAM requirements for different quantization methods and context sizes

About Qwen3 235B A22B Thinking

The Qwen3-235B-A22B-Thinking model is a specialized reasoning variant within the Qwen3 family, developed by Alibaba. It is engineered specifically for tasks requiring high levels of cognitive processing, such as multi-step logical deduction, complex mathematical proofs, and advanced scientific analysis. As a causal language model, it differs from general-purpose models by being permanently optimized for a reasoning-first approach. It generates internal chain-of-thought traces, typically encapsulated within system-defined thinking blocks, to maintain transparency and maximize accuracy in problem-solving environments.

Architecturally, the model utilizes a sparse Mixture-of-Experts (MoE) transformer framework, consisting of 128 total experts. During any single inference pass, the routing mechanism dynamically selects and activates 8 experts per token, resulting in approximately 22 billion active parameters from a total pool of 235 billion. This design provides the representational capacity of a massive parameter space while maintaining the computational profile and latency characteristic of a smaller dense model. The system further incorporates Grouped-Query Attention (GQA) with a 64:4 head ratio and 94 transformer layers, balancing high-throughput inference with modeling of long-range dependencies.

Technical performance is supported by a native context window of 262,144 tokens, facilitating the processing of extensive documents and complex agentic workflows. To ensure stability during large-scale deployments, the model employs RMSNorm for normalization and the SwiGLU activation function. For position encoding, it utilizes Rotary Positional Embeddings (RoPE), which allow for generalization to varying sequence lengths. This iteration represents an enhanced version of the Qwen3 reasoning architecture, featuring refined training on step-by-step analytical datasets to improve performance in coding, STEM, and strategic planning domains.

Technical Specifications

Resources

About Qwen 3

The Alibaba Qwen 3 model family comprises dense and Mixture-of-Experts (MoE) architectures, with parameter counts from 0.6B to 235B. Key innovations include a hybrid reasoning system, offering 'thinking' and 'non-thinking' modes for adaptive processing, and support for extensive context windows, enhancing efficiency and scalability.

Other Qwen 3 Models

• Qwen3-0.6B
• Qwen3-1.7B
• Qwen3-4B
• Qwen3-8B
• Qwen3-14B
• Qwen3-30B-A3B
• Qwen3-32B
• Qwen3 Next 80B A3B
• Qwen3-235B-A22B
• Qwen3 Coder 480B A35B