intent.preference: "Speed" and the sweep is tuned for response time over marginal price improvement: the fastest-answering LPs are prioritised, timeouts tighten, and ranking favors quotes that arrive early. Auction-driven and liquidation-style integrations use this to get an executable answer in as little as ~100 ms from the nearest region.
There’s a faster tier still: a dedicated low-latency lane quoting exclusively from the quickest LPs, switched on per workspace by the TetraFi team. If your flow needs it, raise it with support - regional endpoints, solverOptions shortlists, and tightened sweep timeouts are all part of that conversation.