Skip to main content
Oqoqo helps you build evals and custom benchmarks for real work. Define the tasks that matter to your product. Run coding agents against those tasks in realistic cloud environments. Compare agents, models, and the interfaces you give them — skills, MCP servers, CLIs, and SDKs. Read pass rates, lift, traces, and frictions.

​
What you can do

  • Measure how well agents use your product
  • Build custom benchmarks from your own task sets
  • Compare agents and models for your use cases
  • Find frictions in interfaces and token waste
  • Work from the dashboard, Ask Oqo, MCP, or the CLI
Oqoqo dashboard with its main navigation areas.

​
Main terms

See Core concepts for the attachment model. See the Glossary for full definitions.

​
Ways to work

MCP, CLI, and Ask Oqo share the same product capabilities. Secrets stay in the dashboard.

​
Basic workflow

1

Set up

Connect model providers. Enable agents. Assign Oqoqo AI features.
2

Define

Write a task and rubric. Attach files and a machine to the task. Attach skills or tools to treatments.
3

Launch

Choose agents, treatments, and trials. Oqoqo runs each combination in the cloud.
4

Measure

Read Matrix, Frictions, and run detail. Fix the interface or product. Launch again.

​
Next steps

Quickstart

Launch your first experiment.

Supported agents

See which agents run today.

Model providers

Connect keys, gateways, and subscriptions.

Connect with MCP

Install the plugin, then let your coding agent operate Oqoqo.
Organizations start on the free plan, which resets runs each month. No credit card is required. See Pricing for current plan details. Subscribe or top up from Organization settings → Usage & billing.