Snek - Lightning-Fast AI Code Completion

Write code, not prompts.

Open-source AI autocomplete powered by Cerebras. Lightning-fast completions where you control the context.

🆓 100% Free & Open Source ⚡ Powered by Cerebras

> Blazing fast completions (1800+ tokens/sec)

> Context-aware suggestions that understand your project

> 100% open source - no telemetry, no account required

> Works with VSCode, Neovim, and any LSP-compatible editor

> Powered by Cerebras for unmatched speed and latency

> Why Cerebras?

Snek exclusively uses Cerebras for the best possible UX

⚡ 1,800+ Tokens/Second

10x faster than traditional GPU inference. Completions appear instantly as you type.

🎯 Sub-100ms Latency

No laggy, stuttering suggestions. Snek feels like it's reading your mind.

✨ Best UX

The speed makes suggestions feel like part of your natural coding rhythm.

💰 Free Tier Available

Cerebras offers a generous free tier to get you started.

Note: Snek currently only supports Cerebras. While the architecture could support other providers, we've optimized exclusively for Cerebras's speed to deliver the best user experience.

> How It Works

Three steps to context-aware coding

Step 1: Share Context with Snek

Fill Snek's context with relevant data

> Write your ideas and guides into markdown files and put into .snek/{sessionID}/context/ folder
> Put relevant code snippets into .snek/{sessionID}/code_snippets.json
> Alternatively use the /snek.share command with Claude Code or other AI assistants

Step 2: Snek Watches

Snek reads the context files and code snippets you shared. It builds a complete understanding of:

> Your intent on what you want to make
> Relationships between different files/functions/code logic components
> Function/object signatures and patterns
> Your coding conventions and project architecture

Step 3: Snek Suggests

Start coding naturally. Snek suggests completions based on the full context:

> Press Tab to accept suggestions
> Keep typing to ignore them
> Every suggestion is context-aware and lightning-fast
> You stay in complete control

> Get Started

Setup takes less than 5 minutes

Step 1: Get a Cerebras API Key (Free)

Visit cloud.cerebras.ai
Sign up for a free account
Create an API key
Copy your API key

Cerebras offers a generous free tier to get you started!

Step 2: Install Snek

For VSCode:

Install from VSCode Marketplace
Open Settings (⌘, or Ctrl+,)
Search for snek.apiKey
Paste your Cerebras API key
(Optional) Choose a model in snek.model

For Neovim:

See the snek-nvim repository for installation instructions.

For Other Editors:

Snek works with any LSP-compatible editor. See the Integration Guide.

Step 3: Start Coding!

That's it! Open any supported file (Rust, Python, JavaScript, TypeScript, Go, C/C++, Java, Lua) and start coding.

Snek will automatically create a .snek/ directory in your project and start providing completions.

> Frequently Asked Questions

Is Snek really free?

Yes! Snek is 100% free and open source under the MIT license. You only need a Cerebras API key, which has a generous free tier. There are no subscriptions, no limits, and no hidden costs for the software itself.

Why Cerebras only? Can I use other providers?

Snek is optimized for Cerebras because of its unmatched speed (1800+ tokens/sec) and low latency (sub-100ms). This makes completions feel instant and natural. While the architecture could support other providers, we've chosen to focus on Cerebras to deliver the best possible user experience.

How is Snek different from Copilot or Cursor?

Snek is not an AI coding agent. It's designed for developers who want to write their own code but need intelligent, context-aware suggestions. Unlike Copilot, you control exactly what context Snek uses (via markdown files and code snippets). Unlike Cursor, Snek is open source, runs locally, and has no telemetry.

Does Snek work with my programming language?

Snek supports Rust, Python, JavaScript, TypeScript, Go, C/C++, Java, and Lua. The context-aware engine works across languages and understands polyglot projects.

What data does Snek send to external services?

Snek only sends your code context (the file you're editing + context files you've explicitly added) to Cerebras for completions. No telemetry, analytics, or usage data is collected. All context stays on your machine. The code is open source, so you can verify this yourself.

Can I contribute to Snek?

Absolutely! Snek is open source and we welcome contributions. Check out the GitHub repository to report bugs, request features, or submit pull requests.

Which Cerebras model should I use?

Start with llama3.1-8b (default) for the best balance of speed and quality. If you need better context understanding, try llama3.1-70b or llama-3.3-70b, but they're slower. The 8B model is plenty fast for most use cases.