verial benchmarks commands manage Benchmarks — ordered groups of tasks that reference an environment and are evaluated against criteria. Use them to script benchmark definitions alongside your agent code.
Authentication is required. See verial auth.
Subcommands
| Command | Description |
|---|---|
verial benchmarks list | List benchmarks in your organization. |
verial benchmarks create | Create a new benchmark. |
verial benchmarks get | Get a single benchmark by ID. |
verial benchmarks update | Update benchmark metadata or run limits. |
verial benchmarks delete | Delete a benchmark. |
verial benchmarks list
List benchmarks visible to your organization. Synopsis:verial benchmarks create
Create a benchmark. Tasks and criteria are attached through the REST API or dashboard after creation. Synopsis:| Flag | Description | Default |
|---|---|---|
--name <name> | Benchmark name. | Required |
--description <desc> | Free-text description. | — |
--timeout <seconds> | Per-task timeout in seconds. | 300 |
--concurrency <n> | Max tasks to run in parallel during a run. | 1 |
verial benchmarks get
Fetch a single benchmark, including tasks and criteria. Synopsis:verial benchmarks update
Update benchmark metadata or run limits. Omitted fields are left unchanged. Synopsis:verial benchmarks delete
Delete a benchmark. Previously completed runs are retained. Synopsis:Next Steps
Benchmarks
Benchmark model, tasks, and criteria.
Benchmarks API
Full REST parameters and response schema.