Qwen
Qwen/Qwen3-Coder-Next

OverviewRuns

Quantizations

QuantQuantized bySizeDecodePrefillScoreActions
4bitMLX's logo.
MLX Community
MLX Community		41.8 GB47.1 tok/s648.5 tok/sRuns poorly
q4_k_mGGML's logo.
Unsloth
Unsloth		45.2 GBN/AN/AN/A

Device Comparison

Results include trials with 4,096 input tokens and 1,024 output tokens only.

Decode / Prefill Speeds

3 devices