Pro Search Retrieval Benchmark & Cost Matrix

Workbench v2.6
Recommended Architecture
Highest multi-hop precision efficiency
Retrieval Accuracy Gain
+47%
Vs Standard Single-Hop RAG
Cost Efficiency Ratio
3.42x
Precision per total USD spent
Total Simulated Cost
$0.0384
Across 4 agentic hops
Interactive Multi-Hop Query Tree Graph (D3.js Propagation)
Root Payload
High Relevance (>0.80)
Medium Snippet
Pruned Noise
Multi-Model Performance & Cost Benchmark Matrix
Model Architecture Avg Latency (ms) In / Out Tokens API + Search Cost Precision Score Noise Ratio Efficiency Status
Multi-Hop Diagnostic Telemetry Log
workbench_proof_surface
RECOMMENDED MODEL Kimi K2.6 (Multi-Step Agent)
RETRIEVAL ACCURACY GAIN +47%
COST EFFICIENCY RATIO 3.42
SIMULATED TOTAL COST $0.0384
LATENCY PENALTY 1850 ms
Enjoy this tool? Build your own with Super