Last updated 07-30-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
Lora
Lora is a local large language model SDK from Peekaboo Labs that runs inside Flutter mobile apps on iOS and Android. The 2.4 billion parameter model weighs about 1.5GB and processes prompts entirely on the device, so no internet connection or cloud API is needed. Developers integrate it through the lora_sdk Flutter package with a license key from Peekaboo Labs.
Cloud LLM APIs send every user prompt to a remote server. Lora keeps inference on the phone, which means no data leaves the device and no per-token API bill. Peekaboo Labs reports performance comparable to GPT-4o-mini while using less memory and battery than larger models. The trade-off is model capacity: Lora targets mobile chat and assistant features, not open-ended research tasks.
Flutter developers building privacy-sensitive apps are the core audience. That includes health trackers, offline assistants, and enterprise tools in air-gapped environments. The SDK supports streaming text generation, model download progress tracking, and built-in error handling. Flutter 3.0+, Dart 3.0+, iOS 13+, and Android API 23+ are required.
2.4 billion parameter model at 1.5GB size optimized for mobile devices
lora_sdk Flutter package integrates with a single initialize call
Runs fully offline with no internet connection required
Supports iOS 13+ and Android API 23+ through Flutter 3.0+
Streaming text generation with model download and warmup progress tracking
Performance comparable to GPT-4o-mini per Peekaboo Labs benchmarks
Runs fully offline with no cloud API dependency
Lightweight 1.5GB model sized for mobile hardware
Single-line SDK initialization in Flutter projects
No user data sent to external servers during inference
Flutter-only with no native iOS or Android SDK
Requires a paid license key to operate in production
2.4B parameters limit capability vs larger cloud models
What is Lora by Peekaboo Labs?
Lora is a local large language model SDK for Flutter mobile apps. Lora runs a 2.4B parameter model on the device so prompts never leave the phone, making it suitable for offline and privacy-sensitive applications.
Does Lora require an internet connection?
No, Lora runs entirely on the device after the model is downloaded. Lora works in airplane mode with no cloud API calls, which keeps user data private and eliminates per-token API costs.
How do I integrate Lora into a Flutter app?
Add the lora_sdk package from pub.dev to your Flutter project, obtain a license key from Peekaboo Labs, and call LoraSdk.initialize with your key and an LLM provider. A test app repo is available on GitHub for reference.
What platforms does Lora support?
Lora supports iOS 13 and above and Android with minSdk 23 through the Flutter framework. Lora targets Flutter only and does not offer native Swift or Kotlin SDKs.
How big is the Lora model?
Lora uses a 2.4 billion parameter model that occupies about 1.5GB on the device. Peekaboo Labs optimized Lora for lower energy use and faster inference compared to larger cloud models.
How much does Lora cost?
Lora uses a license key model with a Starter plan reported at $99 per month for unlimited tokens and one app. Lora also offers custom Enterprise pricing with multi-app support and one-on-one technical help.