> ## Documentation Index > Fetch the complete documentation index at: https://developers.flameproxies.com/llms.txt > Use this file to discover all available pages before exploring further. # Proxy Speed & Latency > What affects proxy latency, what to expect from each proxy type, and how to measure and improve performance in your setup. Every proxied request takes a longer path than a direct one. This page explains where the time goes, what you can expect from each proxy type, and how to make your setup faster. ## What latency means in a proxy setup A proxied request makes several hops: 1. Your client connects to the FlameProxies gateway. 2. The gateway forwards the request to a residential IP. 3. That IP connects to the target website. 4. The response travels back the same way. Each hop adds time. The biggest variable is usually step 3: residential IPs are real home connections, and their speed and distance from the target vary. ## What to expect by proxy type | Proxy type | Speed | | - | - | | Residential | Depends on the home connection behind each IP. Use fast mode to route through the lowest-latency peer. | | Premium residential | Same gateway and modes as residential, drawn from a curated pool of clean IPs. | | Static residential (ISP) | The fastest option: dedicated IPs on enterprise hardware, with port speeds up to 10 Gbps. | ## What affects latency in practice * **Distance.** The further the exit IP is from the target's servers, the longer each round trip takes. A request exiting in Brazil to a server in Germany will be slower than one exiting in Germany. * **Targeting precision.** Narrow targeting, like a small city, leaves fewer IPs to choose from, and the gateway can't always pick a fast one. * **New connections.** Every new connection needs a TCP and TLS handshake through the proxy. Rotating on every request means paying that cost every time. * **Page weight.** Large pages with images, video, and scripts take longer to transfer and use more bandwidth. * **The target.** A slow or rate-limiting website is slow through any proxy. ## How to reduce latency * **Turn on fast mode.** Add `mode-fast` to route through the peer with the lowest latency available. Fast mode doesn't cost extra bandwidth. ```text theme={null} -package-standard-country-us-mode-fast ``` * **Try each pool.** Sub-pools have different performance characteristics. Compare `pool-1` and `pool-2` against your target. * **Exit near the target.** When location doesn't matter for your data, target the country where the website's servers are. * **Reuse connections for multi-step work.** For flows that make several requests, use a sticky session and let your client keep the connection open, so you only pay the handshake once. * **Load less.** In headless browsers, block images, fonts, and media you don't need. * **Use static IPs for throughput.** For large transfers to a fixed set of targets, [static residential proxies](/proxies/static-residential) are much faster and have unlimited bandwidth. ## How to measure Use curl's timing output to see where the time goes: ```bash theme={null} curl -x proxy.flameproxies.com:8989 -U "USERNAME-package-standard-country-us:PASSWORD" \ -o /dev/null -s \ -w "connect: %{time_connect}s tls: %{time_appconnect}s first byte: %{time_starttransfer}s total: %{time_total}s\n" \ https://ipinfo.io/json ``` * `connect` is the time to reach the gateway. * `tls` is when the TLS handshake with the target finished, through the proxy. * `first byte` is when the target started responding. * `total` is the whole request. Run it several times — and with and without `mode-fast` — to compare. A single request tells you little, because each rotating request uses a different IP. ## Next steps Pools, modes, and every parameter. Dedicated IPs with up to 10 Gbps port speed.