Introduction to MIND

MIND (Machine Intelligence Native Design) is a systems programming language for AI development.

Key Features

tensor<f32[3]>

Native Tensor Operations

First-class support for tensors.

@differentiable

Differentiable Programming

Built-in automatic differentiation.

  • x86
  • ARM

Hardware Aware

x86 and ARM CPU backends ship today with cross-substrate bit-identity (integer + Q16.16); GPU/accelerator execution is on the commercial roadmap.

Code Example

fn main() {
    // Dense tensor literal: Phase 11 Tier 1 (v0.10.0)
    let t: tensor<f32[3]> = [1.0, 2.0, 3.0];
    // Elementwise ops work in-function
    let y = t .* 2.0;
    print(y);
}

Feature Status

  • Shipped

    Self-hosting

    Pure-MIND front-end emits native-ELF byte-identically (v0.10.0); the native-ELF self-host fixed point is closed

  • Partial

    Tensors

    Dense literals and elementwise ops Phase 11 Tier 1; inter-fn tensor args and intrinsics on roadmap

  • Shipped, scoped

    Determinism

    i64 and Q16.16 byte-identical cross-substrate; scalar f64/f32 run-to-run bit-identical on the strict path (verified on x86_64 (AVX2) + ARM64 (NEON), 2026-07-05); vector-reduction float reproducible within-substrate

  • Partial

    Autodiff

    Compile-time gradients for scalars and core ops; full tensor AD is Phase 12

  • Primitives shipped

    Crypto & protocol primitives

    Pure-MIND AES-128-GCM, SHA-256, HKDF, X25519, X.509 parse+verify, SHA-3/SHAKE, RSA-PSS, ECDSA-P256, ML-KEM-768, TLS 1.3 (key schedule/record/handshake), HPACK, and HTTP/2 framing, RFC/NIST-KAT-verified as a primitive library; a socket-driven TLS client/server is roadmap