List your company

Mathematical Reasoning on AGIEval-MATH (test)

93.3AccuracyAgent 1
23.82841.86459.977.936Jan 30, 2026Feb 15, 2026Mar 4, 2026Mar 20, 2026Apr 6, 2026Apr 22, 2026May 9, 2026

Evaluation results

MethodLinks
Agent 1Controller Backbone=Qwen3-4B-Base, Agent 1 Model=Qwen3-30B-A3B-Instruct2026.0593.3
Agent 1Controller Backbone=Qwen3-8B-Base, Agent 1 Model=Qwen3-30B-A3B-Instruct2026.0593.3
Iterative Critique-and-Routing ControllerController Backbone=Qwen3-8B-Base2026.0589.3
Iterative Critique-and-Routing ControllerController Backbone=Qwen3-4B-Base2026.0587.8
Controller V2Controller Backbone=Qwen3-4B-Base2026.0580.9
Router-R1Controller Backbone=Qwen3-4B-Base2026.0579.8
Controller V2Controller Backbone=Qwen3-8B-Base2026.0577.6
Random RouterController Backbone=Qwen3-4B-Base2026.0573.7
Controller V1Controller Backbone=Qwen3-4B-Base2026.0573.6
RouterDCController Backbone=Qwen3-4B-Base2026.0573.6
Controller V1Controller Backbone=Qwen3-8B-Base2026.0573.5
Agent 2Controller Backbone=Qwen3-4B-Base, Agent 2 Model=Qwen2.5-7B-Instruct2026.0573.4
RoBERTa RouterController Backbone=Qwen3-4B-Base2026.0572.9
Agent 2Controller Backbone=Qwen3-8B-Base, Agent 2 Model=Ministral-3-8B-Instruct2026.0568.1
RouterDCController Backbone=Qwen3-8B-Base2026.0565.4
Random RouterController Backbone=Qwen3-8B-Base2026.0564
Router-R1Controller Backbone=Qwen3-8B-Base2026.0563.8
RoBERTa RouterController Backbone=Qwen3-8B-Base2026.0562.8
Agent 3Controller Backbone=Qwen3-4B-Base, Agent 3 Model=Qwen2.5-1.5B-Instruct2026.0554.7
UPAVenue=-, Execution Engine=GPT-4o-mini2026.0152.1
Step-BackVenue=ICLR 24, Execution Engine=GPT-4o-mini2026.0147.5
OPROVenue=ICLR 24, Execution Engine=GPT-4o-mini2026.0146.1
PromptBreederVenue=ICML 24, Execution Engine=GPT-4o-mini2026.0145.9
SPOVenue=EMNLP 26, Execution Engine=GPT-4o-mini2026.0145.3
CoTVenue=NeurIPS 22, Execution Engine=GPT-4o-mini2026.0144.5
APEVenue=ICLR 23, Execution Engine=GPT-4o-mini2026.0144.4
TextGradVenue=Nature 25, Execution Engine=GPT-4o-mini2026.0144.4
IOVenue=-, Execution Engine=GPT-4o-mini2026.0142.1
RaRVenue=arXiv 23, Execution Engine=GPT-4o-mini2026.0142.1
PromptAgentVenue=ICLR 24, Execution Engine=GPT-4o-mini2026.0141.4
Agent 3Controller Backbone=Qwen3-8B-Base, Agent 3 Model=Llama-3-2-1B-Instruct2026.0526.5