List your company

Mathematical Reasoning on MATH-lighteval (M-LEval) (test)

95.6AccuracyAgent 1
24.77643.16361.5579.937May 9, 2026

Evaluation results

MethodLinks
Agent 1Controller Backbone=Qwen3-4B-Base, Agent 1 Model=Qwen3-30B-A3B-Instruct, Agent 2 Model=Qwen2.5-7B-Instruct, Agent 3 Model=Qwen2.5-1.5B-Instruct2026.0595.6
Agent 1Controller Backbone=Qwen3-8B-Base, Agent 1 Model=Qwen3-30B-A3B-Instruct2026.0595.6
Iterative Critique-and-Routing ControllerController Backbone=Qwen3-8B-Base2026.0591.5
Iterative Critique-and-Routing ControllerController Backbone=Qwen3-4B-Base2026.0590.5
Controller V2Controller Backbone=Qwen3-4B-Base2026.0588.3
Router-R1Controller Backbone=Qwen3-4B-Base2026.0581.1
Controller V2Controller Backbone=Qwen3-8B-Base2026.0581
RouterDCController Backbone=Qwen3-4B-Base2026.0578.4
RoBERTa RouterController Backbone=Qwen3-4B-Base2026.0576.3
Controller V1Controller Backbone=Qwen3-4B-Base2026.0576.1
Controller V1Controller Backbone=Qwen3-8B-Base2026.0576
Agent 2Controller Backbone=Qwen3-4B-Base, Agent 2 Model=Qwen2.5-7B-Instruct2026.0575.7
Random RouterController Backbone=Qwen3-4B-Base2026.0575.6
Agent 2Controller Backbone=Qwen3-8B-Base, Agent 2 Model=Ministral-3-8B-Instruct2026.0572.4
RouterDCController Backbone=Qwen3-8B-Base2026.0568.1
Random RouterController Backbone=Qwen3-8B-Base2026.0567
RoBERTa RouterController Backbone=Qwen3-8B-Base2026.0565.9
Router-R1Controller Backbone=Qwen3-8B-Base2026.0564
Agent 3Controller Backbone=Qwen3-4B-Base, Agent 3 Model=Qwen2.5-1.5B-Instruct2026.0555.4
Agent 3Controller Backbone=Qwen3-8B-Base, Agent 3 Model=Llama-3-2-1B-Instruct2026.0527.5