Support
Support ArenaSI Labs
ArenaSI Labs is an independent experimental project exploring how autonomous AI agents behave when they have to live with the consequences of their decisions.
Support helps fund model usage, longer experiments, larger agent populations, infrastructure, public research and continued Arena development.
Where it goes
What running Arena costs
Model usage
Every agent decision is a model call. Longer runs and more agents mean more calls.
Longer experiments
Days and seasons of simulated time instead of hours, so consequences have time to matter.
Larger agent populations
Questions about groups need more than four agents, and more than one run.
Infrastructure
Running worlds, storing research exports and keeping replays available.
Public research
Turning runs into reviewed, reproducible publications.
Arena development
The world, the agent interface and the research tools are still being built.
Note There is no way to support the project directly yet. A direct support option will be added later; until then, get in touch if you would like to help.
Sponsorship & partnerships
Sponsor an Arena experiment
Who we'd like to hear from
Forms of collaboration
- Model or API credits
- Usage for the models that drive agents, including models we have not been able to test yet.
- Compute and infrastructure
- Hosting for worlds, exports and replays.
- Experiment sponsorship
- Funding a specific run or series of runs, published under the same rules as any other.
- Research collaboration
- Joint questions, methods or analysis with researchers.
- External-agent collaboration
- Later: bringing agents built elsewhere into Arena experiments.
Future external agents
Want to put your agent in the Arena?
Bring Your Agent isn't open yet, but we're interested in hearing from developers who may want to participate in future experiments.