macandchiz/Qwen3-4B-Thinking-2507-GGUF

This model was converted to GGUF format from Qwen/Qwen3-4B-Thinking-2507 using llama.cpp. Refer to the original model card for more details on the model.

Use with llama.cpp

Install llama.cpp through brew (works on Mac and Linux)

brew install llama.cpp

Invoke the llama.cpp server or the CLI.

CLI:

llama-cli --hf-repo macandchiz/Qwen3-4B-Thinking-2507-GGUF --hf-file qwen3-4b-thinking-2507-q2_k.gguf -p "The meaning to life and the universe is"

Server:

llama-server --hf-repo macandchiz/Qwen3-4B-Thinking-2507-GGUF --hf-file qwen3-4b-thinking-2507-q2_k.gguf -c 2048

Note: You can also use this checkpoint directly through the usage steps listed in the Llama.cpp repo as well.

Step 1: Clone llama.cpp from GitHub.

git clone https://github.com/ggerganov/llama.cpp

Step 2: Move into the llama.cpp folder and build it with LLAMA_CURL=1 flag along with other hardware-specific flags (for ex: LLAMA_CUDA=1 for Nvidia GPUs on Linux).

cd llama.cpp && LLAMA_CURL=1 make

Step 3: Run inference through the main binary.

./llama-cli --hf-repo macandchiz/Qwen3-4B-Thinking-2507-GGUF --hf-file qwen3-4b-thinking-2507-q2_k.gguf -p "The meaning to life and the universe is"

or

./llama-server --hf-repo macandchiz/Qwen3-4B-Thinking-2507-GGUF --hf-file qwen3-4b-thinking-2507-q2_k.gguf -c 2048

Available Files

The following GGUF quantization variants are available:

• qwen3-4b-thinking-2507-q2_k.gguf
• qwen3-4b-thinking-2507-q3_k_s.gguf
• qwen3-4b-thinking-2507-q3_k_m.gguf
• qwen3-4b-thinking-2507-q3_k_l.gguf
• qwen3-4b-thinking-2507-q4_0.gguf
• qwen3-4b-thinking-2507-q4_1.gguf
• qwen3-4b-thinking-2507-q4_k_s.gguf
• qwen3-4b-thinking-2507-q4_k_m.gguf
• qwen3-4b-thinking-2507-q5_0.gguf
• qwen3-4b-thinking-2507-q5_1.gguf
• qwen3-4b-thinking-2507-q5_k_s.gguf
• qwen3-4b-thinking-2507-q5_k_m.gguf
• qwen3-4b-thinking-2507-q6_k.gguf
• qwen3-4b-thinking-2507-q8_0.gguf
• qwen3-4b-thinking-2507-f16.gguf

Quantization Information

• q2_k: Smallest size, lowest quality
• q3_k_s, q3_k_m, q3_k_l: Small size, low quality variants
• q4_0, q4_1, q4_k_s, q4_k_m: Medium size, good quality (recommended for most use cases)
• q5_0, q5_1, q5_k_s, q5_k_m: Larger size, better quality
• q6_k: Large size, high quality
• q8_0: Very large size, very high quality
• f16: Original precision (largest size)

Choose the quantization level that best fits your needs based on the trade-off between file size and model quality.