Techub News reports that AI search company Perplexity has released its in-house developed Rust-based retrieval and ranking engine, Photon, to replace its previously forked open-source engine. Photon now handles all production traffic and supports the new "Fast Search" mode in its search API. Perplexity reports that its single-call p50 latency is 160 milliseconds, and p95 latency is 230 milliseconds. Photon has reduced the p99 latency for retrieval and ranking from approximately 800 milliseconds to about 65 milliseconds, while using approximately 20% fewer service machines compared to the old content nodes. The engine is provided as a managed API, and users can access it by setting search_type: "fast" in POST /search, with a cost of $1 per 1000 requests, although the engine itself is not open source. Perplexity has also launched a "Fast Search" mode optimized for agent workflows alongside Photon. In 3554 tasks across 6 benchmarks, this mode achieved a score of 64.3% with an estimated model and search cost of $59.73, while the default preset cost $187.60, making the former about 68% cheaper.(MarkTechPost)
如您认为本页面(或本平台)内容存在侵权或违规情形,请您在发现之日起尽快与我们联系,并提供:权利人姓名/名称及联系方式、被侵权作品名称及权属证明、具体侵权链接及情况说明。我们在核实后将及时处理相关内容。
联系方式:creator@techub.news
商务合作:contacting@techub.news
点击链接下载 Techub News App,获取最新讯息:
https://www.techub.news/download
(或扫描下方二维码下载)

All information provided on this website is for reference only.
This website does not guarantee the accuracy, validity, timeliness, or completeness of the information.
Any actions taken based on the information provided by this website are at the user's own risk.
Techub News reports that AI search company Perplexity has released its in-house developed Rust-based retrieval and ranking engine, Photon, to replace its previously forked open-source engine. Photon now handles all production traffic and supports the new "Fast Search" mode in its search API. Perplexity reports that its single-call p50 latency is 160 milliseconds, and p95 latency is 230 milliseconds. Photon has reduced the p99 latency for retrieval and ranking from approximately 800 milliseconds to about 65 milliseconds, while using approximately 20% fewer service machines compared to the old content nodes. The engine is provided as a managed API, and users can access it by setting search_type: "fast" in POST /search, with a cost of $1 per 1000 requests, although the engine itself is not open source. Perplexity has also launched a "Fast Search" mode optimized for agent workflows alongside Photon. In 3554 tasks across 6 benchmarks, this mode achieved a score of 64.3% with an estimated model and search cost of $59.73, while the default preset cost $187.60, making the former about 68% cheaper.(MarkTechPost)
如您认为本页面(或本平台)内容存在侵权或违规情形,请您在发现之日起尽快与我们联系,并提供:权利人姓名/名称及联系方式、被侵权作品名称及权属证明、具体侵权链接及情况说明。我们在核实后将及时处理相关内容。
联系方式:creator@techub.news
商务合作:contacting@techub.news
点击链接下载 Techub News App,获取最新讯息:
https://www.techub.news/download
(或扫描下方二维码下载)

All information provided on this website is for reference only. The website does not guarantee the accuracy, validity, timeliness, or completeness of the information. Any actions based on the information provided are at the user’s own risk.