pub.dev

ai_edge

0.2.0KyoheiG3/ai_edge ↗New releases in your daily digest, changelog included.

Flutter plugin for on-device AI inference powered by MediaPipe GenAI. Run large language models directly on iOS and Android with optimized performance.

Last release 1 years ago
16 Sep 2025
Too new to tell
only 2 release windows
Nearly every release is documented
notes for 3 of 3 stable releases
Nothing withdrawn
no release was ever pulled
1 years old
3 releases · first in 2025

Release timeline

3 releases since 2025

20252026

One column per month.

Releases

0.2.016 Sept 2025
refactor: simplify API by removing config objects by @KyoheiG3 in #11
What's Changed

refactor: simplify API by removing config objects by @KyoheiG3 in #11

chore: publish packages by @KyoheiG3 in #12

Full Changelog: ai_edge-v0.1.0...ai_edge-v0.2.0
Open source →
Additional notes
Breaking Changes

API Simplification: Replaced config objects with individual named parameters (#11)

createModel() now accepts individual parameters instead of ModelConfig object

createSession() now accepts individual parameters instead of SessionConfig object

initialize() now accepts individual parameters for both model and session configuration

Config classes (ModelConfig and SessionConfig) are still available internally but not exposed in public API

Improvements

Developer Experience:

More intuitive API with direct parameter passing

Better IDE autocomplete support without config object nesting

Optional parameters with sensible defaults:

maxTokens: Default 1024 (previously required)

temperature: Default 0.8

randomSeed: Default 1

topK: Default 40

Platform Implementation:

Made non-critical parameters optional in Android native implementation

Fixed platform channel communication issues with Unit type handling

Code Quality:

Removed unnecessary toString() override from GenerationEvent class

Updated documentation to remove references to deprecated exception types

Updated tests to reflect the new API design

From the same repository

KyoheiG3/ai_edge

0.1.016 Sept 2025

refactor: reorganize project into monorepo structure by @KyoheiG3 in #8

What's Changed

refactor: reorganize project into monorepo structure by @KyoheiG3 in #8
refactor: improve ai_edge package structure and error handling by @KyoheiG3 in #9
chore: prepare ai_edge package v0.1.0 for publishing by @KyoheiG3 in #10

Full Changelog: ai_edge-v0.0.1...ai_edge-v0.1.0

Open source →

Additional notes

Breaking Changes

Package Structure: Reorganized package to use src/ directory structure
- Main implementation moved to lib/src/ directory
- Public API now exported through lib/ai_edge.dart
- Platform interfaces moved to src/ subdirectory
Error Handling: Complete overhaul of error handling system
- Introduced sealed classes (AiEdgeException) for exhaustive error handling
- Replaced string-based errors with type-safe exception hierarchy
- CallError now accepts Exception directly instead of message strings
- Removed unnecessary PlatformException conversions
Platform Interface: Standardized data conversion patterns
- Platform interfaces now accept Map<String, dynamic> instead of typed objects
- Data conversion happens in upper layer (ai_edge.dart) for consistency
- Cleaner separation between platform communication and business logic

Features

Monorepo Structure: Project reorganized as Flutter workspace
- Packages moved to packages/ directory
- Examples moved to examples/ directory
- Enables better package management and development workflow

Improvements

Code Quality
- Improved test coverage with updated mock implementations
- Better separation of concerns with src/ directory structure
- More maintainable and extensible codebase
Developer Experience
- Cleaner public API surface
- Better IDE support with organized file structure
- Improved error messages with sealed class pattern

Chores

Added LICENSE and README.md symlinks for pub.dev publishing requirements
Updated dependencies and build configurations
Improved Android build configuration

Open source →

0.0.18 Sept 2025

feat: streamline integration test infrastructure and CI/CD workflows by @KyoheiG3 in #1

What's Changed

feat: streamline integration test infrastructure and CI/CD workflows by @KyoheiG3 in #1
refactor: relocate test models to project-relative directory by @KyoheiG3 in #2
docs: improve package documentation for pub.dev release by @KyoheiG3 in #3
feat: add example app with model management and chat UI by @KyoheiG3 in #4
style: format all Dart files with consistent style by @KyoheiG3 in #5
ci: add comprehensive CI workflow with platform builds by @KyoheiG3 in #6
chore: publish packages by @KyoheiG3 in #7

New Contributors

@KyoheiG3 made their first contribution in #1

Full Changelog: https://github.com/KyoheiG3/ai_edge/commits/ai_edge-v0.0.1

Open source →

Additional notes

Initial release of the AI Edge Flutter plugin for on-device AI inference using MediaPipe GenAI.

Features

Native AI Inference: On-device AI inference support for both iOS and Android platforms using MediaPipe GenAI
Streaming Responses: Real-time streaming text generation with async stream support
Model Management: Built-in model download and management capabilities
Example App: Comprehensive example application with chat UI and model management interface
Platform Support:
- iOS 13.0+ with MediaPipeTasksGenAI (0.10.24)
- Android with MediaPipe GenAI (0.10.24)

Infrastructure

Comprehensive CI/CD workflow with platform-specific builds
Integration and unit test infrastructure
Automated code formatting and linting
Pull request template for consistent contributions

Developer Experience

Project documentation for pub.dev release
Tool version management with mise configuration
Simplified plugin architecture with clean API design
Example app demonstrating all plugin capabilities

Open source →