OpenDGO
OpenDGO: Benchmarking Multi-Agent Organizations Across Diverse Application Domains
A controlled benchmark evaluating multi-agent organization structures across six domains and three collaboration settings to determine optimal agent coordination strategies.
How to run
$ npm i -g lettokenburn-cli$ ltb bench info opendgo
External harness. No Harbor adapter has been generated for this benchmark yet. `ltb bench info` prints the known links.
Runs happen on your machine with your own subscription; only the score summary is pushed. CLI docs
Scores
1 sourced entryEvery number links to where it was published. Low-confidence rows are muted and need per-cell verification.
Reproduced by the community
All runsAbout
OpenDGO covers Coding, Finance, Biomedical, Legal, Scientific Research, and Recommendation domains under Sequential, Parallel, and Integration settings. It compares fixed and learned organizations by varying communication graphs, roles, and inputs.
The framework analyzes performance trade-offs, showing no single organization dominates all domains. It connects results to graph properties and communication traces to explain effectiveness under specific resource and failure conditions.