CascadeFlow v0.6.0 Released

[saschabuehrle]

CascadeFlow v0.6.0 Release Notes

We're excited to announce CascadeFlow v0.6.0 with production-ready LangChain integration, multi-instance provider support, and access to 200+ models through OpenRouter!

New Features

✨ 100% Python/TypeScript Feature Parity Achieved

Full feature parity between Python and TypeScript implementations across the entire CascadeFlow library.

Highlights:

• Identical API and functionality across both languages
• 1036/1036 TypeScript tests passing (100% core functionality coverage)
• 19 matching examples in both languages
• Complete API documentation with TypeDoc (auto-generated)
• Migration guide for seamless Python → TypeScript transitions
• Batch processing, streaming, tool calling, quality validation, cost tracking all available in both languages

TypeScript-Specific Enhancements:

• Browser usage examples (React, Vue, Webpack, Vite)
• Deno runtime support
• Vercel Edge Functions deployment guide
• Express.js integration patterns

🔗 LangChain Integration (Python & TypeScript)

Production-ready LangChain integration for both Python and TypeScript with zero-config setup.

Features:

• LangChain-compatible callback handlers for cost and token tracking
• Built-in cost tracking utilities with budget management and CSV export
• Automatic LangSmith tag tracking and integration
• Universal provider support across all LangChain models

Documentation:

• LangChain Integration Guide
• Python Configuration
• TypeScript Package README

Examples:

• Python: Basic Usage | Streaming | Cost Tracking | LangSmith | Cascade Benchmark
• TypeScript: Basic Usage | Cross-Provider | Full Benchmark

🖥️ Multi-Instance Provider Support

Run draft and verifier models on separate Ollama/vLLM instances for optimized deployment.

Documentation:

• Multi-Instance Providers Guide

Examples:

• Python: Ollama Multi-Instance | vLLM Multi-Instance
• TypeScript: Ollama Multi-Instance | vLLM Multi-Instance

🌐 OpenRouter Provider

Access 200+ models through unified OpenRouter integration.

Documentation:

• OpenRouter Provider Guide

Bug Fixes

• Critical n8n integration fixes: Resolved cascading issues and improved reliability for n8n workflow automation
• Fixed multi-instance providers to correctly support per-model base_url configuration
• Corrected API usage in multi-instance Ollama and vLLM examples
• Fixed o1-mini model ID (o1-mini-2024-09-12) and documented Tier 5 requirement
• Resolved AttributeError in edge_device.py example
• Fixed Python code quality issues for Ruff linting compliance (deprecated type hints, unused imports)

Installation

Python:

pip install cascadeflow

TypeScript/Node.js:

npm install @cascadeflow/core

LangChain Integration:

# Python
pip install cascadeflow

# TypeScript
npm install langchain-cascadeflow

---

Full changelog: https://github.com/lemony-ai/cascadeflow/releases/tag/v0.6.0