Vector database that can index 1B vectors in 48M

I was starting to think this was impressive, if not impossible. 1B vectors in 48 MB of storage => Maybe not impossible using shared/lossy storage if they were sparsely scattered over a large space ?

But anyways - minutes. Thanks.

Edit: Gemini suggested that this sort of (lossy) storage size could be achieved using "Product Quantization" (sub vectors, clustering, cluster indices), giving an example of 256 dimensional vectors being stored at an average of 6 bits per vector, with ANN being one application that might use this.

Re: Vector database that can index 1B vectors in 48M

#6

Post by l5870uoo9y » Fri, Sep 12, 2025, 6:07 PM UTC

ge96 wrote: post #2

M is minutes

Thankfully not months.

Re: Vector database that can index 1B vectors in 48M

#7

Post by softwaredoug » Fri, Sep 12, 2025, 6:07 PM UTC

Not trying to be snarky, just curious -- How is this different from TurboPuffer and other serverless, object storage backed vector DBs?

Re: Vector database that can index 1B vectors in 48M

#8

Post by softwaredoug » Fri, Sep 12, 2025, 6:08 PM UTC

l5870uoo9y wrote: post #6

ge96 wrote: post #2

M is minutes

Thankfully not months.

Oh the horrors of search indexing Ive seen... including weeks / months to rebuild an index.

Re: Vector database that can index 1B vectors in 48M

#9

Post by ashvardanian » Fri, Sep 12, 2025, 6:14 PM UTC

Very curious about the hardware setup used for this benchmark!

Re: Vector database that can index 1B vectors in 48M

#10

Post by CuriouslyC » Fri, Sep 12, 2025, 6:19 PM UTC

OutOfHere wrote: post #4

Proprietary closed-source lock-in. Nothing to see here.

Seriously. The amount of lift a SaaS product needs to give me is insane for me to even bother evaluating it, and there's a near zero percent chance I'll use it in my core.