Qwen3-1.7B-Base-4bit (MLX)
Qwen/Qwen3-1.7B-Base converted to MLX with 4-bit
quantization (mlx_lm.convert -q --q-bits 4), unchanged otherwise. It is
the model behind the AI picks and completions of
海酷拼音 Haiku Pinyin on the Mac,
which downloads it from here.
The weights are Qwen's, under the Apache License 2.0 (LICENSE).