Set `speed: 'fast'` once; gateway routes to low-latency tier when available, falls back to standard automatically.
Summary
Eliminates provider-specific fast mode syntax—single parameter works across all models on AI Gateway. Reduces latency/throughput tradeoffs without manual routing logic or model pinning.
Why it matters
Eliminates provider-specific fast mode syntax—single parameter works across all models on AI Gateway. Reduces latency/throughput tradeoffs without manual routing logic or model pinning.
Implementation verdict
Replaces per-provider fast mode APIs with unified gateway abstraction. Requires updating `providerOptions.gateway` parameter in existing `generateText` calls. Ready now (beta); fast variants cost more per token but no adoption friction.
Sources
Dev Signal
Get briefs like this in your inbox — free, every weekday.
100+ sources compressed into one 4-minute read. Ranked, cited, implementation-ready.