BenchFlow

Trust & data

How the vendor protects buyer material, where its data comes from, and where it can run.

Deployment

Vendor-hosted

Source:posttrain.com

Data sources

Product replica or partner app

Source:benchflow.ai

Quote

• Product replica or partner app: ClawsBench is five high-fidelity mock workplaces — claw-gmail, claw-gcal, claw-gdocs, claw-gdrive, claw-slack — wire-compatible with the upstream Google Workspace `gws` CLI and the Slack Web API.

Stated confidentiality practices

Its runtime hardens verifier workflows and runs agents as non-root sandbox users to limit tampering with evaluation materials; no customer-data confidentiality controls are publicly stated.

Source:benchflow.ai

Not publicly listed

SOC 2, Trust center, Other certifications, Exclusivity

What they sell

Offerings

RL environmentsEvals and benchmarksRL infrastructure

Sources:benchflow.aigithub.com

Domains

Multi-domainEnterprise workflows

Sources:benchflow.ai/research/skillsbenchbenchflow.ai/research/clawsbench

Delivery formats

Docker containersAPIDatasetsManaged service

Sources:benchflow.ai/docs/benchflow/sandboxesbenchflow.ai/docs/benchflowbenchflow.ai/research/clawsbenchposttrain.com

Public samples & benchmarks

• SkillsBench
• ClawsBench
• PostTrain Arena
• FrontierPhysics

Sources:benchflow.ai/research/skillsbenchbenchflow.ai/research/clawsbenchbenchflow.ai/research

Named customers

Not confirmed

Sources checked:benchflow.ai