Fold

50-78%

Token Reduction

<1ms

Compression Time

Zero

API Calls Required

Quick Example

import { fold } from "@fold/sdk";

const ctx = fold();  // That's it!

ctx.system("You are a coding assistant...");
ctx.think("I need to search for information...");
ctx.act({ tool: "search", query: "..." }, "search");
ctx.observe("Found 3 results...", "search");

// Get optimized messages for your LLM
const messages = ctx.messages();

// Check your savings
console.log(ctx.saved());
// { tokens: 5000, percent: 45, cost: 0.05 }

Why Fold?

Massive Savings

Cut token costs by 50-78% without sacrificing task performance. Based on NeurIPS 2025 research.

Sub-millisecond

Zero API calls for compression. Masking happens locally in under 1ms.

Semantic Preservation

Not just truncation. Fold understands your context and preserves what matters.

Drop-in Integration

Works with OpenAI, Anthropic, Vercel AI SDK, and LangChain out of the box.

Get Started

Quick Start Install Fold and start saving tokens in under 5 minutes.

Core Concepts Learn about sessions, blocks, and masking strategies.

Integrations Use Fold with your favorite frameworks.

API Reference Complete documentation for all exports.

Framework Support

Framework	Integration
OpenAI SDK	foldMessages(), wrapOpenAI()
Anthropic SDK	foldAnthropicMessages()
Vercel AI SDK	withFold(), useFold()
LangChain/LangGraph	ContextManagedMemory

Ready to cut your LLM costs?

Get started with Fold in under 5 minutes.

Get Started