welcome to octra
user and developer documentation for the octra network
documentation is updated continuously as the network grows and evolves, and some materials may become momentarily outdated. always double-check octra labs’ official github.
octra is inherently a collaborative project. if you have any suggestions, you are welcome to send a pull request or join the discussion in our telegram or in the discord development channel.
for specific technical questions, write to: [email protected]
what is octra
octra is an FHE blockchain network with support for isolated execution environments. it can function both as a standalone L1 blockchain and as decentralized encrypted middleware for existing networks and applications.
octra was founded in 2021 and has been in active development since late 2022. an internal prototype was released in october 2023, a public testnet was launched in june 2025, and mainnet alpha was launched in december 2025.
the majority of octra’s code is written in OCaml and C++. octra’s public codebase and tooling are progressively open-sourced through official repositories.
what is FHE
fully homomorphic encryption, or FHE, allows computation on encrypted data without decrypting it.
in the context of octra, this means that the network can process all kinds of data, execute arbitrary logic, and return results, all while the underlying data remains encrypted.
the FHE scheme that is used in octra network is called HFHE, or Hypergraph Fully Homomorphic Encryption. it is an in-house development based on hypergraphs and designed to enable consumer-grade applications with the help of parallel encrypted computations.
some octra libraries can be easily ported to other projects. for example, the HFHE library can be compatible with any project and can operate independently.
how octra works
octra combines a blockchain network, a HFHE-based encrypted compute layer, and isolated execution environments called circles.
circles can host programs with dedicated application logic and encrypted data storage. programs can be built in native AppliedML, as well as Rust, C++, OCaml, or WASM.
octra accepts all kinds of data, which is transformed into vectors, then encrypted and processed using HFHE. encryption and decryption require R1CS proof generation.
the network automatically distributes storage and validation between nodes of three types: bootstrap nodes, standard nodes, and light nodes.
further information can be found in the octra network litepaper (2024).
octra virtual machine
OVM supports AppliedML as the native language for programs. it also supports a diverse array of languages through WASM. an IDE is built-in for the current version of the web client.
Last updated a day ago