Deep Research Agent Topology & Model Routing Workbench

Pipeline Settings

Architecture Preset

Query Fan-Out Branching: 4

Max Hop Search Depth: 3

Chunk Size (Tokens): 1000

Vector Prune Threshold: 0.75

Input Cost ($/M tok): $0.60

Output Cost ($/M tok): $2.40

Execution Telemetry

Estimated Cost

$0.1656

Estimated Latency

14.2s

Input Tokens

142,500

Output Tokens

8,400

Context Window Fill 71.3% (142.5k / 200k)

Reasoning & Grounding Index

High Reasoning / Optimal Grounding

12 queries executed • 48 text chunks processed

Trace Log

[Ready] Pipeline initialized with Kimi K2.6 preset.

[Ready] Click 'Simulate Execution' to run state path.