Benchmark·Public sample

BankerToolBench

A banking-workflow benchmark used to evaluate agents and verifier behavior on artifact-heavy tasks.

ConfirmedConfidence: medium·Last verified: 2026-08-16·How verification works
Access
Public / Open
Profile completeness 100%

What it is

BankerToolBench contains banking-domain tasks whose outputs can include spreadsheets and other work products. Handshake uses a 21-task slice in its reported BankerVerifierBench meta-evaluation of grading systems.

Capabilities

Only evidenced capabilities are shown as Yes. Missing evidence remains Unknown.

StatefulUnknown
Long horizonUnknown
Multi-turnUnknown
Multi-agentUnknown
AdversarialUnknown
MultimodalUnknown
Tool useYes
Browser useUnknown
Computer useUnknown
Code executionUnknown
File manipulationYes
API interactionUnknown
MCP supportUnknown
SandboxedUnknown
Deterministic resetUnknown
Stochastic behaviorUnknown
Human in the loopUnknown
Expert authoredYes
SyntheticUnknown
Production derivedUnknown
Real application replicaUnknown
Partial observabilityUnknown
Persistent stateUnknown
Cross-applicationUnknown

Sources & provenance

  1. Handshake AI Research · dataset · accessed 2026-08-16 · confirmed
  2. Handshake AI Research · official · accessed 2026-08-16 · company-reported

Related artifacts