GRPO (Mixture-of-experts routing): superseded — cited as a baseline and beaten by newer methods. 1 paper(s) critique it, 2 beat it on benchmarks — #60 of 1370 most-superseded. Sub-problem: cluster led by Mixtral. Newer alternatives in the same sub-problem include PADD, MESA, PR2, RouteHijack, MASCing.

Mixture-of-experts routing

GRPO

Superseded baseline#60 of 1,370 most-superseded

Superseded — cited as a baseline and beaten by newer methods

1 papers critique it · 2 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites GRPO as a baseline.

GRPO can suffer from abrupt reward collapse on MoE models.

— Towards Stable and Effective Reinforcement Learning for Mixture-of-Experts

Beaten on benchmarks

Head-to-head results where a newer method reports beating GRPO. Values are copied from the source paper's tables — verify against the cited paper.

GMPO+RS (RSPO) beats GRPO
85.2 vs 70.7
Pass@1 · [Code]
Towards Stable and Effective Reinforcement Learning for Mixture-of-Experts
MESA beats GRPO
96.44 vs 96.36
GSM8K · [Model: Qwen3-30B-A3B]
MESA: Improving MoE Safety Alignment via Decentralized Expertise

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.