Show HN: OpenWebSearch – A router for web search indexes

Show HN: OpenWebSearch – A router for web search indexes(openwebsearch.ai)

3 points by yoeven 45 days ago | 3 comments

Hi HN, I'm one of the people behind OpenWebSearch (https://openwebsearch.ai).

It's a router for web search indexes. You POST to one endpoint with a `provider` field, and it normalizes both the request and the response for web search indexes like Parallel, Brave, Exa and more

Why we built it: we run a model company (Interfaze) and a lot of our models are smaller in size and we're experimenting if given web search can a smaller 9b or 70b model perform the same as 300b or 600b model and we found that it does extremely better when given web search similar to this paper (https://arxiv.org/abs/2203.05115)

but we also found not all web search are built the same, some are better in people search, some better at financial data and others are bio research, etc.

Like LLMs, web indexes are becoming commoditized with different indexes having different strengths and weaknesses with access to niche data, performance and cost. Every large model lab including Interfaze has to build their own internal mini-Google for training and eventually launch that index as a service.

Some cool features: - Centralized billing - Standardized input and output structure - Fallback support if a provider goes down - Cost tracking

Full blog: https://interfaze.ai/blog/introducing-openwebsearch

kesku 45 days ago |

this is neat! two thoughts though:

1. adding another hop in front of the search provider adds latency, which as an inference company I imagine you know sucks

2. I'm not convinced you actually want to route between this many search providers, for inference you generally just want the best web search available, which obviously I'd argue is Exa :) I can understand something like Apify + Exa where you need specialized/unique data, but otherwise the routing feels like extra complexity without much upside?

yoeven 45 days ago | |

thanks!

1. yeah, we've optimized significantly for latency, give the API a try and you'll see it's pretty fast.

2. more than you think, we use a wide range of providers internally depending on the context needed. Exa could be great in some for example people search but is pretty bad at getting the latest content of the page, Parallel is better there. Apify is great for getting google ranked pages. Perplexity is great at financial data which Exa is pretty bad at. Context7 (coming soon) great at code, way better than exa for code. Depending on the use case, different indexes do better. we're also seeing companies with access to prosperity data like Valyu.

gauravguitara 40 days ago |

Can you expose the traces over api?