FHErnet

$FHERNET is live 0x34204508568e49e8ff4710b481fb5f10b4a74f52 Fully Homomorphic Encryption allows computations to be performed on encrypted data. Here’s how private search works in FHErnet: You describe the model you need. Your browser creates an embedding locally. The embedding is encrypted before leaving your device. The server calculates similarity scores using the encrypted vector. Your browser decrypts the scores and reveals the best matches. You can search for “a cheap model for debugging large codebases” without sending the raw query or an unencrypted embedding to the search server.

Need more context? Care about price? Keep those preferences private. FHErnet ranks the catalog with encrypted weights. Your browser decrypts the scores and shows your matches. fhernet.site/private-routing

Choose a model. Send a prompt. Inspect a real response. FHErnet brings model discovery and inference into one interface. FHE protects private discovery; the selected inference provider can read the prompt. fhernet.site

Find a model without sharing your search intent. FHErnet creates an embedding locally, computes similarity on encrypted vectors, and decrypts the matches in your browser. Private discovery starts here: fhernet.site/private-search

Find models within your budget without sharing the limit in plaintext. FHErnet compares public prices against your encrypted budget. Your browser decrypts the differences and filters the matches.

Private discovery. Live inference. Now available on InferDAG. Discover a model privately, inspect its capabilities and turn a coding prompt into a working response all through FHErnet. Model discovery stays encrypted. Inference prompts remain visible to the provider you choose.

Encrypted model routing demo Your model preferences shouldn't have to be public. Set your priorities. FHErnet encrypts them, computes encrypted ranking scores, and decrypts the results locally. Private discovery. A ranking you can inspect.

New feature is live on InferDAG: Private Model Discovery Find the right model without revealing your search query. FHErnet creates the embedding locally, calculates similarity across encrypted vectors and decrypts the results in your browser. Your search. Your intent. Your privacy.

Find a model that fits the task. Explore FHErnet’s marketplace, filter by price, and compare model context windows, availability, and pricing evidence side by side.

We just launched Encrypted Budget Matching on FHErnet. Change your budget and watch affordable options appear without exposing your limit or price differences. Computation stays encrypted, only the final results are decrypted and filtered locally.

One prompt. Two models. A clear quote before payment. Compare models, prepare inference requests, and review the provider’s verified price in FHErnet.

We just shipped another new feature on FHErnet : Marketplace to quote From browsing models to reviewing an inference quote. Filter the marketplace → compare options → prepare your prompt → check the price before payment.

Discover models without exposing your search query or budget. 🔐 FHErnet computes on encrypted inputs. Your browser decrypts the results and filters budget matches locally. Private discovery, powered by FHE.

The server can calculate a match without reading your query vector. That’s FHE in action: compute on encrypted data, return encrypted scores, decrypt locally.

Your budget shouldn’t be public. Here is the demo of our new feature Private Budge. FHErnet now matches models to an encrypted spending limit. 1. Encrypt your budget locally 2. Compute price differences on encrypted data 3. Decrypt and filter matches in your browser Try it: fhernet.site/private-routing…

Just added another feature in FHErnet Added and deployed Private budget matching. You can encrypt your budget locally. It calculate encrypted price differences on the server. Also, decrypts and filters matches in your browser. Posting demo next.