Build-Bench
Competition dates

Timeline

The participant schedule from the website beta and invited pilot through public evaluation, the final Agent freeze, and publication of the official results.

Competition schedule

Use these dates to plan Agent development, testing, and final-version selection. Operational limits published on August 31 will apply when public development and validation open.

  1. Release

    Website beta and initial documentation

    The participant-facing website beta, updated dataset and Starter Kit information, and initial competition documentation are released. Teams can begin reviewing the task, submission format, and local development workflow.

  2. Pilot

    Invited pilot

    Invited teams test registration, Agent upload, Hosted Smoke Test, Full Evaluation, and the supporting infrastructure. Pilot feedback may be used to correct the interface before the public phase.

  3. Policy freeze

    Rules and resource limits frozen; registration opens

    The organizers publish the frozen dataset split sizes, per-Case timeouts, submission limits, resource policy, build-bypass policy, and validation protocol. Team registration opens to eligible participants on the same date.

  4. Opening

    Public development and validation open

    Public development resources, Agent submission, Hosted Smoke Test, and public validation open to eligible teams.

  5. Deadline

    Final Agent version freeze

    Each Team must select the qualified, immutable Agent version to be used for final hidden evaluation. Later uploads will not replace the selected final version.

  6. Results

    Final results published

    Hidden evaluation and result review conclude, and the official final results are published by this date.

Schedule updates. If a date changes, this page will be versioned and the change will be announced through the official competition website.