Skip to main content

Run this benchmark in OrgX

OrgX Autonomous Initiative Benchmark

The easiest way to evaluate OrgX is to run the benchmark inside OrgX itself. We provision the benchmark workspace, execute the suite across every OrgX agent domain, and give you a scorecard, surfaced artifacts, and a public-style proof bundle at the end.

15–25 min

Starter benchmark

A fast, all-domain proof run that exercises the benchmark pipeline and returns a scorecard, surfaced artifacts, and a benchmark bundle.

60–90 min

Full benchmark week

The full public suite with repeats, cross-domain coverage, benchmark facts, and the same publication-grade output shape used for weekly reporting.

Sign up to run the benchmarkAlready have an account? Open Benchmark LabRead the benchmark specInspect the public repo

Benchmark flow

  1. 1. Sign up for OrgX or open Benchmark Lab if you already have access.
  2. 2. Run the starter benchmark for a fast all-domain proof pass.
  3. 3. Open your benchmark run detail page to inspect scorecards and artifacts.
  4. 4. Compare your run to the public weekly benchmark corpus.

Latest public week

232.17x flow multiplier

Week local-openai-gpt-5-nano-full-public-judge-20260411 · 15 tasks · cross_functional, design, engineering, marketing, ops, product, sales

View the benchmark bundle

Run it in OrgX

No local install, no fixture wrangling, no benchmark scripts. Sign up, open Benchmark Lab, and run the benchmark inside OrgX itself.

Compare three execution modes

Every benchmark task is measured against OrgX orchestration, a strong single-agent baseline, and a curated human baseline.

Inspect the proof

Weekly scorecards, methodology, raw public bundles, and the public benchmark repo stay available for inspection and citation.