Qwen3-235B-A22B-GGUF

Qwen3 Highlights

Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models. Built upon extensive training, Qwen3 delivers groundbreaking advancements in reasoning, instruction-following, agent capabilities, and multilingual support, with the following key features:

• Uniquely support of seamless switching between thinking mode (for complex logical reasoning, math, and coding) and non-thinking mode (for efficient, general-purpose dialogue) within single model, ensuring optimal performance across various scenarios.
• Significantly enhancement in its reasoning capabilities, surpassing previous QwQ (in thinking mode) and Qwen2.5 instruct models (in non-thinking mode) on mathematics, code generation, and commonsense logical reasoning.
• Superior human preference alignment, excelling in creative writing, role-playing, multi-turn dialogues, and instruction following, to deliver a more natural, engaging, and immersive conversational experience.
• Expertise in agent capabilities, enabling precise integration with external tools in both thinking and unthinking modes and achieving leading performance among open-source models in complex agent-based tasks.
• Support of 100+ languages and dialects with strong capabilities for multilingual instruction following and translation.

Model Overview

Qwen3-235B-A22B has the following features:

• Type: Causal Language Models

• Training Stage: Pretraining & Post-training

• Number of Parameters: 235B in total and 22B activated

• Number of Paramaters (Non-Embedding): 234B

• Number of Layers: 94

• Number of Attention Heads (GQA): 64 for Q and 4 for KV

• Number of Experts: 128

• Number of Activated Experts: 8

• Context Length: 32,768 natively and 131,072 tokens with YaRN.

• Quantization: q4_K_M, q5_0, q5_K_M, q6_K, q8_0

For more details, including benchmark evaluation, hardware requirements, and inference performance, please refer to our blog, GitHub, and Documentation.

Quickstart

llama.cpp

Check out our llama.cpp documentation for more usage guide.

We advise you to clone llama.cpp and install it following the official guide. We follow the latest version of llama.cpp. In the following demonstration, we assume that you are running commands under the repository llama.cpp.

Cloning the repo may be inefficient, and thus you can manually download the GGUF file that you need or use huggingface-cli (pip install huggingface_hub) as shown below:

huggingface-cli download Qwen/Qwen3-235B-A22B-GGUF Q4_K_M/Qwen3-235B-A22B-Q4_K_M-00001-of-00005.gguf --local-dir . --local-dir-use-symlinks False

However, for large files, we split them into multiple segments due to the limitation of 50G for a single file to be uploaded. Specifically, for the split files, they share a prefix, with a suffix indicating its index. For examples, the Q4_K_M GGUF files are:

Qwen3-235B-A22B-Q4_K_M-00001-of-00005.gguf
Qwen3-235B-A22B-Q4_K_M-00002-of-00005.gguf
Qwen3-235B-A22B-Q4_K_M-00003-of-00005.gguf
Qwen3-235B-A22B-Q4_K_M-00004-of-00005.gguf
Qwen3-235B-A22B-Q4_K_M-00005-of-00005.gguf

They share the prefix of Qwen3-235B-A22B-Q4_K_M, but have their own suffix for indexing respectively, say -00001-of-00005. To use the split GGUF files, you need to merge them first with the command llama-gguf-split as shown below:

./llama-gguf-split --merge Qwen3-235B-A22B-Q4_K_M-00001-of-00005.gguf Qwen3-235B-A22B-Q4_K_M.gguf

Switching Between Thinking and Non-Thinking Mode

You can add /think and /no_think to user prompts or system messages to switch the model's thinking mode from turn to turn. The model will follow the most recent instruction in multi-turn conversations.

Here is an example of multi-turn conversation:

> Who are you /no_think

<think>

</think>

I am Qwen, a large-scale language model developed by Alibaba Cloud. [...]

> How many 'r's are in 'strawberries'? /think

<think>
Okay, let's see. The user is asking how many times the letter 'r' appears in the word "strawberries". [...]
</think>

The word strawberries contains 3 instances of the letter r. [...]

Processing Long Texts

Qwen3 natively supports context lengths of up to 32,768 tokens. For conversations where the total length (including both input and output) significantly exceeds this limit, we recommend using RoPE scaling techniques to handle long texts effectively. We have validated the model's performance on context lengths of up to 131,072 tokens using the YaRN method.

To enable YARN in llama.cpp:

./llama-cli ... -c 131072 --rope-scaling yarn --rope-scale 4 --yarn-orig-ctx 32768

[!NOTE] All the notable open-source frameworks implement static YaRN, which means the scaling factor remains constant regardless of input length, potentially impacting performance on shorter texts. We advise adding the rope_scaling configuration only when processing long contexts is required. It is also recommended to modify the factor as needed. For example, if the typical context length for your application is 65,536 tokens, it would be better to set factor as 2.0.

[!TIP] The endpoint provided by Alibaba Model Studio supports dynamic YaRN by default and no extra configuration is needed.

Best Practices

To achieve optimal performance, we recommend the following settings:

1. Sampling Parameters:

   • For thinking mode (enable_thinking=True), use Temperature=0.6, TopP=0.95, TopK=20, MinP=0, and PresencePenalty=1.5. DO NOT use greedy decoding, as it can lead to performance degradation and endless repetitions.
   • For non-thinking mode (enable_thinking=False), we suggest using Temperature=0.7, TopP=0.8, TopK=20, MinP=0, and PresencePenalty=1.5.
   • We recommend setting presence_penalty to 1.5 for quantized models to suppress repetitive outputs. You can adjust the presence_penalty parameter between 0 and 2. A higher value may occasionally lead to language mixing and a slight reduction in model performance.

2. Adequate Output Length: We recommend using an output length of 32,768 tokens for most queries. For benchmarking on highly complex problems, such as those found in math and programming competitions, we suggest setting the max output length to 38,912 tokens. This provides the model with sufficient space to generate detailed and comprehensive responses, thereby enhancing its overall performance.

3. Standardize Output Format: We recommend using prompts to standardize model outputs when benchmarking.

   • Math Problems: Include "Please reason step by step, and put your final answer within \boxed{}." in the prompt.
   • Multiple-Choice Questions: Add the following JSON structure to the prompt to standardize responses: "Please show your choice in the answer field with only the choice letter, e.g., "answer": "C"."

4. No Thinking Content in History: In multi-turn conversations, the historical model output should only include the final output part and does not need to include the thinking content. It is implemented in the provided chat template in Jinja2. However, for frameworks that do not directly use the Jinja2 chat template, it is up to the developers to ensure that the best practice is followed.

Citation

If you find our work helpful, feel free to give us a cite.

@misc{qwen3technicalreport,
      title={Qwen3 Technical Report}, 
      author={Qwen Team},
      year={2025},
      eprint={2505.09388},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2505.09388}, 
}