Qwen2.5-Coder-0.5B

Qwen2.5-Coder-0.5B is the smallest member of Alibaba's Qwen2.5-Coder family, released November 11, 2024, alongside five larger siblings from 1.5B to 32B. It brings the same Apache 2.0 licensing and code-focused training to a footprint small enough for edge deployment.

Despite its size, the model handles code generation, completion, debugging, and explanation with a 32K context window, enough to process sizable files or multiple modules at once. The 0.5B instruct variant is specifically aimed at local IDE integrations, on-device debugging tools, and real-time code analysis where a cloud round-trip would add unacceptable latency.

Weights are open under Apache 2.0, and the community has produced GPTQ quantized builds in both 8-bit and 4-bit precision that run with near-zero VRAM overhead. The larger 32B sibling in the same release was reported to match GPT-4-class coding models on HumanEval and MBPP.