YanahCoder model files (Qwen3-Coder-Next, streamed-experts layout)
Files for the YanahCoder Mac app, which streams the mixture-of-experts weights from SSD instead of holding them in RAM.
dense/: the non-expert weights of a 4-bit MLX conversion (attention, Gated DeltaNet, shared experts, embeddings, routers), the tokenizer and config.experts-4bit/: routed experts (4-bit), one record per expert, split into one file per layer.yanah_manifest.json: size and sha256 of every file.
Derived from Qwen/Qwen3-Coder-Next (Apache-2.0). Quantization reduces quality versus the original model.
Hardware compatibility
Log In to add your hardware
Quantized
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for Giovanniwants/YanahCoder-Qwen3-Coder-Next
Base model
Qwen/Qwen3-Coder-Next