Nicheloom

Market intelligence for builders — see what's gaining traction before it's crowded.

Cachet

A drop-in semantic cache for LLM APIs, 100% local, in Rust

Details

External ID
48643854
Source
HN
Company
—
Product
Cachet
Website domain
github.com
Launched
June 23, 2026
Cohort
—
Upvotes
5
Upvotes percentile
0.12568306010928962
Tags
—
Fetched at
Sept. 7, 2026, 9:26 p.m.
Updated at
Sept. 7, 2026, 9:26 p.m.

Enrichment

Theme
ML inference and model optimization
Vertical
Horizontal
Function
Model & infra
Audience
Developer
AI stance
AI feature
Project type
Commercial product
Normalized one-liner
semantic cache for llm api calls
Manually corrected
False

Could you build this?

Partial While an HTTP proxy that hashes and caches LLM outputs is simple, building an ultra-fast, local semantic cache in Rust requires embedding models, fast vector indexing, and approximate nearest neighbor search.

What it would actually take: The implementation relies on Rust (using Axum or Actix) embedding an ONNX runtime (via ort or candle) to run local embedding models (like all-MiniLM-L6-v2) alongside a vector index (like HNSW or USearch) and an embedded key-value store (like RocksDB or Sled). The difficult aspects are optimizing inference and cosine distance searches to achieve sub-millisecond overhead while preserving high semantic recall and managing memory bounds. It requires systems programming in Rust and familiarity with vector search internals.

Discussion

No comments on this launch.

Competitors

Other products that read as similar to this one — 1624 launches clear the similarity bar, closest 8 shown.

Attention rank: #1348 of 1625 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).

Launched 237 days after the earliest competitor.

Other launches for this product

Same idea, different domain

Nobody's really built a model & infra tool for Fintech yet.