OpenLanguage dropped on Hacker News this week as a free, open-source conversational language tutor built specifically for iOS. The project pairs an AI tutor with native voice I/O capabilities, letting users practice speaking and listening without the typical walled-garden approach that dominates language learning apps today. No accounts, no subscription tiers, no premium features locked behind paywalls—just plug in your own API key and start talking.

Built on Modern iOS Stack

The application is constructed using Expo and React Native, leveraging native SwiftUI components through expo-ui for a genuinely Apple-native feel rather than the clunky cross-platform compromises many mobile apps settle for. Under the hood, it taps into iOS's built-in speech recognition and text-to-speech systems, meaning voice capabilities work out of the box without additional API costs or dependencies. This architectural choice keeps the app lightweight while maintaining professional-grade accessibility features.

Zero Telemetry by Design

What sets OpenLanguage apart from competitors like Duolingo or Rosetta Stone isn't just the price—it's the fundamental privacy posture. The project explicitly states zero telemetry: no analytics, no tracking, no data collection of any kind. Conversations stay entirely on-device and flow only to whatever API provider you've configured. For users concerned about ed-tech companies monetizing their learning patterns and speaking mistakes, this represents a meaningful philosophical departure from the surveillance capitalism model that's become industry standard.

Bring Your Own Keys: The Economics Work Differently

Rather than charging subscription fees for AI access, OpenLanguage employs a bring-your-own-keys (BYOK) model. Users supply their own API credentials from OpenAI, Anthropic, or any compatible provider—including DeepSeek, Fireworks, Groq, Mistral, Moonshot, TogetherAI, Zhipu AI, MiniMax, and several others via the OpenRouter integration. The ai-sdk powers these integrations, making provider expansion straightforward for contributors. This shifts the cost structure dramatically: instead of paying a language app's margin on top of AI costs, users pay only what the model actually charges—often fractions of a cent per conversation session.

Getting Started Requires Apple Hardware

Development setup demands macOS with Xcode and pnpm installed. Running pnpm install followed by pnpm ios spins up the dev server and launches the iOS simulator—no Android support is mentioned, keeping scope tight for what appears to be a small team or solo effort. The project operates under MIT licensing, inviting community contributions through issues and pull requests.

Key Takeaways

  • Completely free and open-source with no monetization layer beyond your own API costs
  • Supports 12+ AI providers through ai-sdk integration for maximum flexibility
  • Voice I/O uses iOS native capabilities at zero additional cost
  • Zero telemetry means learning data never leaves the device except to your chosen provider

The Bottom Line

OpenLanguage is exactly the kind of project that should exist more in the ed-tech space—tooling that respects users enough to let them own their infrastructure. Whether it gains traction against established players with marketing budgets remains to be seen, but for privacy-conscious learners and developers who want to inspect or modify how their language tutor works, this is a refreshing change from the norm.