arXiv: REAM: Merging Improves Pruning of Experts in LLMs
Qwen3-Coder-Next-REAM
This model is a compressed version of Qwen/Qwen3-Coder-Next. It is obtained by reducing the number of experts in each MoE layer from 512 to 384. This reduction is achieved by the REAM method described in https://bknyaz.github.io/blog/2026/moe/.
Compared to other models obtained in this collection, more code data is used in the calibration data during pruning/merging to better preserve original's model coding abilities. Specifically, the ratio between c4, math and coding data (see https://bknyaz.github.io/blog/2026/moe/) is 0.0, 0.3, 0.7. The calibration data used here is the same as in Qwen3-Coder-Next-REAP. Compared to other REAM models, here we used C=32 (number of experts in groups) instead of C=16, which we found to work better.
The compressed model has 60B params (120GB) instead of 80B (160GB) of the original model, reducing storage and GPU memory requirements by roughly 25%. At the same time, the model retains 100% (or very close) of the original model's performance on a variety of benchmarks (see Results section below). Additional efficiency optimization (e.g., quantization) can be added similarly to the original model.
See additional details at Qwen3-30B-A3B-Instruct-2507-REAM.
Results
| Model | IFeval | AIME25 | GSM8K | GPQA-D | HumanEval | LiveCodeBench | AVG |
|---|---|---|---|---|---|---|---|
| Qwen3-Coder-Next | 89.6 | 80.0 | 85.4 | 42.4 | 92.7 | 47.5 | 72.9 |
| Qwen3-Coder-Next-REAM | 89.3 | 80.0 | 85.3 | 40.4 | 94.5 | 48.0 | 72.9 |
License
Please refer to the license of the original model Qwen/Qwen3-Coder-Next.