scann
by google
Anisotropic vector quantization for efficient similarity search
google/scannmixpeek://vector_index@v1/google_scann_v1Overview
ScaNN (Scalable nearest Neighbors) is Google's vector similarity search library optimized for maximum inner-product search on high-dimensional embeddings. It introduces anisotropic vector quantization that outperformed other ANN libraries by 2x on standard benchmarks.
On Mixpeek, ScaNN provides an alternative vector indexing backend with particularly strong performance on inner-product similarity workloads.
Architecture
Uses anisotropic vector quantization — a novel compression technique that significantly boosts accuracy of approximate distance computations compared to standard quantization. Supports asymmetric hashing and tree-based partitioning.
Mixpeek SDK Integration
import { Mixpeek } from "mixpeek";
const mx = new Mixpeek({ apiKey: "API_KEY" });
// Managed: create a collection over a bucket; Mixpeek runs this model's extractor
const collection = await mx.collections.create({
namespace_id: "my-namespace",
collection_name: "my-collection",
source: { type: "bucket", bucket_ids: ["bkt_your_bucket"] },
feature_extractor: {
feature_extractor_name: "image_embedding",
version: "v1",
parameters: { model_id: "MODEL_ID" },
},
});Capabilities
- 2x faster than competing ANN libraries (at time of release)
- Anisotropic vector quantization for better accuracy
- Optimized for maximum inner-product search (MIPS)
- Scales to billion-vector datasets
Use Cases on Mixpeek
Benchmarks
| Dataset | Metric | Score | Source |
|---|---|---|---|
| glove-100-angular (1M) | Recall@10 at 95% target | 99.1% | ann-benchmarks.com |
Performance
Anisotropic vector quantization for better recall-speed tradeoffs
Common Pipeline Companions
Explore on Mixpeek
Compare alternatives in this category
Hand-picked tools & platforms compared
Deep-dive technical guide
See how Mixpeek runs models as extractors
Store & search embeddings at scale
Usage-based pricing for pipelines
Compare models, APIs & infrastructure
Specification
Build a pipeline with scann
Add this model to a processing pipeline alongside other extractors. Combine with retrieval stages for end-to-end search.
Run on your data, free