RL.VENDORS
Chakra Labs logo

Chakra Labs


About

Builds high-fidelity RL environments, evaluations, and human computer-use trajectory datasets for frontier AI research and post-training.

What They Offer

Commercial or operational services this company provides.

Agent evaluations
Rubrics / verifiers
Open-source tooling
Custom RL environments
Training datasets
Computer-use trajectories
Agent trajectories
Managed infrastructure

Products & Public Artifacts

Trajectory Library
Dataset · Access by request

Dataset of 2,500+ hours of human computer-use trajectories, including ~25 billion segmented image tokens and 10 million screenshot/action pairs.

Dojo
Environment · Access by request

Open platform of interactive software clones (Notion, Figma, Canva, Gmail, Slack, and others) for training and evaluating computer-use and tool-use agents, with native support for Harbor, Verifiers, and Verl.

Visit artifact →
MAGI Benchmark
Benchmark · Public

Enterprise benchmark of 1,000 ambiguous, long-horizon, cross-application tasks cloned across 8 real workplace tools; frontier models complete fewer than 30% of tasks at pass@1.

Visit artifact →
Dojo Single-Page Applications (dojo-spas)
Repository · Open source

Mocked single-page application clones used for evals and RL training.

Visit artifact →

Technical Capabilities

Real-App / Workplace SimulationMultimodalLong-Horizon TasksStateful / Persistent EnvironmentsTool / API / MCP UseComputer UseBrowser UseProgrammatic VerifiersHuman / Expert FeedbackSandboxed Execution

Only publicly documented capabilities are shown. Absence does not imply the capability is unavailable.

Focus Areas

Computer UseRL EnvironmentsTraining DataRLHF / Post-trainingEvaluationsHuman Data

Domains

Computer UseEnterprise

Leadership

Alexander Fung
Co-Founder / CTO
Nirmal Krishnan
Co-Founder / CEO