Providers ·
Models ·
Routing Policies ·
Health
Costs
Per-role / per-agent
Routing Test
Where Neco gets model access. Secrets never travel to the browser.
loading…
Every model Neco can route to. Use + Add model to register one manually or pull from the provider's upstream catalog inline.
loading…
Priority chains per task type. The first model in each chain is the primary choice; the rest are fallbacks if it's down or rate-limited. Reorder = changes which model actually answers.
loading…
Per-model availability rolled up by role + provider. Sources: live probes + operator overrides. Detailed latency percentiles + override admin on legacy Model Health.
loading…
Spend tracking. Today / 7-day totals plus per-model and per-tier breakdowns. Daily trend chart + run-level drill-down on legacy Model Costs.
loading…
Which model serves which role inside Neco's brain. Roles map to functions (primary chat, sub-agents, compaction, heartbeat, memory search, summarization); agents inherit these defaults unless they override via their own model_policy. For per-intent routing (chat-vs-coding etc.) see legacy Routing; per-agent overrides land on the Agent detail page.
loading…
What would Neco actually pick? Type a sample message and watch the flow from user message → detected complexity → selected policy → final model.