What you can try
- AFM Base Model: chat against an OpenAI-compatible AFM endpoint.
- ROUTER Decision Routing: the same chat UX through
aria-router, with strategy choices for Semantic / Agent / Decision Model. Gold virtual models:ariacompute/semantic-autoariacompute/agent-autoariacompute/afm-d-auto
- AGENT: sessions against the cloud agent runtime (
aria-agent-cloud). - MEMO memory storage: try the sandboxed
aria-memoinstance. Write tiered memories, run hybridsearch(with an optional graph-aware mode and an inspectableRetrieveTrace), pure-vectorrecall, and batchsearch-batch, then explore the four-view relation plane withconnect/relate/relations/graph. A JSON/text toggle lets you compare parsed results with native CLI output. See the aria-memo CLI guide for command details.
Account
Some Playground actions need a signed-in Aria Compute account on the same site (.com or .cn). See Authentication for sign-in, and Sites & regions for per-site wallets and API keys.