List your company

Long-context Reasoning on LongBench V2 (DocMath, Frames, LB-MQA Test)

66.7DocMath ScoreQwen3-30B-A3B-Thinking + SPELL
14.07627.73841.455.062Sep 28, 2025

Evaluation results

MethodLinks
Qwen3-30B-A3B-Thinking + SPELLModel Category=Reasoning Models, Input Length=100K2025.0966.768.178.450.565.9
Qwen3-30B-A3B-Instruct + SPELLModel Category=Instruct Models, Input Length=100K2025.0964.963.774.848.763
Qwen3-4B-Thinking + SPELLModel Category=Reasoning Models, Input Length=100K2025.0964.860.672.44360.2
Qwen3-30B-A3B-Instruct + RLVRModel Category=Instruct Models, Input Length=100K2025.09646272.447.461.5
Qwen3-30B-A3B-Thinking + RLVRModel Category=Reasoning Models, Input Length=100K2025.0963.967.177.249.664.5
Qwen3-30B-A3B-ThinkingModel Category=Reasoning Models, Input Length=100K2025.0963.865.877.946.763.6
Qwen2.5-32B-Instruct + SPELLModel Category=Instruct Models, Input Length=100K2025.0963.36274.140.860.1
Qwen3-4B-Thinking + RLVRModel Category=Reasoning Models, Input Length=100K2025.0963.358.671.143.459.1
Qwen2.5-32B-InstructModel Category=Instruct Models, Input Length=100K2025.096349.461.536.252.5
Qwen3-30B-A3B-InstructModel Category=Instruct Models, Input Length=100K2025.096357.870.344.158.8
Qwen3-4B-ThinkingModel Category=Reasoning Models, Input Length=100K2025.0961.459.270.940.758.1
R1-Distill-Qwen-14B + SPELLModel Category=Reasoning Models, Input Length=100K2025.0961.162.875.737.959.4
R1-Distill-Qwen-14B + RLVRModel Category=Reasoning Models, Input Length=100K2025.096163.87635.959.2
Qwen2.5-32B + SPELLModel Category=Base Models, Input Length=100K2025.0960.652.262.334.352.4
Qwen2.5-14B-Instruct + SPELLModel Category=Instruct Models, Input Length=100K2025.0960.163.974.840.159.7
Qwen2.5-32B-Instruct + RLVRModel Category=Instruct Models, Input Length=100K2025.0959.762.369.636.957.1
R1-Distill-Qwen-14BModel Category=Reasoning Models, Input Length=100K2025.0959.560.673.333.356.7
Qwen2.5-32B + RLVRModel Category=Base Models, Input Length=100K2025.0957.549.960.132.750.1
Qwen2.5-14B + SPELLModel Category=Base Models, Input Length=100K2025.0956.85363.231.251.1
Qwen2.5-14B-InstructModel Category=Instruct Models, Input Length=100K2025.0956.752.464.236.652.5
Qwen2.5-14B-Instruct + RLVRModel Category=Instruct Models, Input Length=100K2025.0956.759.973.438.557.1
Qwen2.5-14B + RLVRModel Category=Base Models, Input Length=100K2025.0953.252.164.230.550
R1-Distill-Llama-8B + SPELLModel Category=Reasoning Models, Input Length=100K2025.0949.254.37029.350.7
R1-Distill-Llama-8B + RLVRModel Category=Reasoning Models, Input Length=100K2025.0945.4546828.348.9
Qwen2.5-7B-Instruct + SPELLModel Category=Instruct Models, Input Length=100K2025.0944.548.260.732.446.5
Qwen2.5-7B-Instruct + RLVRModel Category=Instruct Models, Input Length=100K2025.0944.148.657.428.244.6
Qwen2.5-7B + RLVRModel Category=Base Models, Input Length=100K2025.0942.740.349.22639.6
R1-Distill-Llama-8BModel Category=Reasoning Models, Input Length=100K2025.0941.552.669.326.447.5
Qwen2.5-32BModel Category=Base Models, Input Length=100K2025.0940.742.250.128.740.4
Qwen2.5-7B + SPELLModel Category=Base Models, Input Length=100K2025.0939.940.150.828.239.8
Llama3.1-8B-Instruct + SPELLModel Category=Instruct Models, Input Length=100K2025.0939.750.860.926.244.4
Qwen2.5-7B-InstructModel Category=Instruct Models, Input Length=100K2025.0939.441.444.528.438.4
Llama3.1-8B-Instruct + RLVRModel Category=Instruct Models, Input Length=100K2025.0936.947.657.226.142
Qwen2.5-14BModel Category=Base Models, Input Length=100K2025.0936.237.543.327.536.1
Llama3.1-8B-InstructModel Category=Instruct Models, Input Length=100K2025.0934.947.353.527.140.7
Qwen2.5-7BModel Category=Base Models, Input Length=100K2025.0916.124.231.222.723.6