The Ternary Intelligence Stack (TIS) is an innovative, sparse AI framework designed to enable efficient frontier intelligence without requiring large-scale infrastructure typical of hyperscalers. Built around balanced ternary logic, this project provides a robust set of tools and a programming language ideal for advanced AI development.

Core Components

1. Ternlang

Ternlang is a systems programming language featuring a unique compiler and inference runtime based on balanced ternary logic. The fundamental type in Ternlang is the trit, allowing for three states: −1 (reject), 0 (hold), and +1 (affirm). This structure introduces deterministic uncertainty, eliminating the problem of missing values. Examples of usage include:

# Install Ternlang CLI
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y && source "$HOME/.cargo/env" && cargo install ternlang-cli

# Run an interactive REPL
ternlang

# Execute a .tern file
ternlang my_program.tern

2. albert.

albert. serves as a ternary Mixture-of-Experts language model, trained completely from scratch using ternary weights ({−γ, 0, +γ}), ensuring high efficiency. This model autonomously adapts its architecture through Net2Net surgery, enhancing performance as it learns. A notable feature is the @sparseskip primitive, which optimizes computation by skipping zero-weight operations, achieving speeds of 83 tokens per second on CPUs. Here’s how to interact with the API:

# API request example (requires an API key)
curl -s https://ternlang-api.fly.dev/api/trit_decide \
  -H "Content-Type: application/json" \
  -H "X-Ternlang-Key: <your-key>" \
  -d '{"statement": "This architecture is worth funding"}' | jq .

3. Agent Albert

Agent Albert is a terminal-native AI agent created in pure Rust. It operates locally without cloud dependencies, providing complete privacy and control over data. Users can initiate this tool easily with:

# Install and launch Agent Albert CLI
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y && source "$HOME/.cargo/env" && cargo install albert-cli

# Launch the agent
albert-cli

Documentation and Resources

Comprehensive documentation is available, including technical specifications, performance metrics, and compliance notes:

• Full Documentation
• Developer Dashboard (Ternlang Studio)
• Architecture & Training Status (Model Card)

The Ternary Intelligence Stack aims to redefine AI stack efficiency, enabling advanced capabilities while remaining accessible without extensive infrastructure needs.