List your company

Tool-augmented agent execution on MCP-Bench

53.5Task FulfillmentGround Truth Server
-1.6896812.6384126.966541.29459May 9, 2026May 15, 2026May 21, 2026May 27, 2026Jun 2, 2026Jun 8, 2026Jun 15, 2026

Evaluation results

MethodLinks
Ground Truth ServerSetting=Default2026.0653.572.674.674.462.355.4——————73.8
SINGSetting=Global2026.0650.567.167.46960.835.3——————67.5
SINGSetting=Restricted2026.0650.166.871.468.360.133.4——————68.4
MCP-ZeroSetting=Restricted2026.0649.562.968.968.358.734.9——————67.6
MCP-ZeroSetting=Global2026.0648.261.1686857.733——————66.7
ReActWorld Model=None2026.0546.836.641.531.240.519.877.7041.736.430.138.936.8
ReAct-Plan-ExecWorld Model=GPT-OSS-120B2026.0538.538.742.547.836.920.610075.438.645.228.887.750
SPIRAL-ExecWorld Model=CLAUDE-SONNET2026.0534.547.555.358.433.327.510079.94156.930.489.954.6
ReAct-Plan-ExecWorld Model=AWM-4B2026.0533.742.645.240.429.626.810085.838.242.828.292.950.5
SPIRAL-ExecWorld Model=GPT-OSS-120B2026.0532.85159.66133.630.910082.841.960.332.291.456.5
SPIRAL-ExecWorld Model=AWM-4B2026.0531.626.633.648.420.520.210010029.14120.310047.6
ReAct-Plan-ExecWorld Model=CLAUDE-SONNET2026.0531.447.253.165.927.129.910078.439.359.528.589.254.1
Qwen2.5-4BMetric Label=Valid ReAct2026.062.43.583.423.782.481.45——2.993.61.97—2.85
Qwen2.5-4BMetric Label=Valid Best2026.061.972.323.173.751.431.38——2.143.461.41—2.34
SmolLM2-3BMetric Label=Valid Best2026.061.482.182.442.511.411.07——1.832.481.24—1.85
Llama-3.2-3BMetric Label=Valid Best2026.061.322.432.622.851.171.1——1.882.731.13—1.91
Llama-3.2-3BMetric Label=Valid ReAct2026.061.251.651.821.821.450.95——1.451.821.2—1.49
DeepSeek-R1-Distill-Qwen-1.5BMetric Label=Valid Best2026.061.051.331.021.1611——1.191.091—1
Qwen2.5-4BMetric Label=SFT2026.060.9461.532.012.951.050.929——1.242.480.987—1.57
Qwen2.5-4BMetric Label=SFT DPO2026.060.9331.372.332.480.90.817——1.152.410.858—1.47
SmolLM2-3BMetric Label=SFT DPO2026.060.8171.050.9831.130.7670.75——0.9331.060.758—0.917
Llama-3.2-3BMetric Label=SFT2026.060.7671.451.20.80.80.717——1.1110.758—0.956
Qwen2.5-4BMetric Label=Valid Init2026.060.7371.351.091.60.6710.554——1.051.350.612—1
DeepSeek-R1-Distill-Qwen-1.5BMetric Label=Valid Init2026.060.7270.8550.7090.8910.6730.673——0.7910.80.673—0.692
Llama-3.2-3BMetric Label=SFT DPO2026.060.7171.021.281.230.7330.75——0.8671.260.742—0.956
SmolLM2-3BMetric Label=SFT2026.060.6750.7710.9381.70.6380.638——0.7231.320.638—0.892
SmolLM2-3BMetric Label=Valid Init2026.060.650.8670.6170.6330.5830.567——0.7580.6250.575—0.653
DeepSeek-R1-Distill-Qwen-1.5BMetric Label=SFT DPO2026.060.60.7330.5670.5670.5670.567——0.6670.5670.567—0.6
SmolLM2-3BMetric Label=Valid ReAct2026.060.5670.550.550.550.550.55——0.5580.550.55—0.553
Llama-3.2-3BMetric Label=Valid Init2026.060.550.850.9170.8670.5830.583——0.70.8920.583—0.725
DeepSeek-R1-Distill-Qwen-1.5BMetric Label=SFT2026.060.4581.280.6250.6080.4580.458——0.8710.6170.458—0.649
DeepSeek-R1-Distill-Qwen-1.5BMetric Label=Valid ReAct2026.060.4330.60.450.4330.4330.433——0.5170.4420.433—0.464