Changelog
All notable changes to Veriafy are documented here
Legend:featureimprovementfixsecuritybreaking
p
v2.3.1
- Fixed memory leak in batch processing on Windows
- Resolved timeout issues with large video files
- Improved error messages for invalid API keys
m
v2.3.0
- Added WebP and AVIF image format support
- New streaming API for real-time classification
- 20% faster document vectorization
- Fixed race condition in concurrent batch requests
m
v2.2.0
- Marketplace launch with 150+ models
- Creator monetization with 80% revenue share
- Model versioning and rollback support
- Enhanced CLI with interactive model browser
m
v2.1.0
- GPU acceleration for 10x faster inference
- Apple Silicon (M1/M2/M3) native support
- AMD ROCm support for Linux
- Reduced memory footprint by 40%
M
v2.0.0
Major Release- Complete rewrite of inference engine (3x faster)
- New base models with improved accuracy
- HEIC/HEIF and MKV/WebM support
- Multi-language document classification
- API v1 deprecated, v2 now default
- Minimum Python version now 3.9
m
v1.9.0
- TypeScript SDK release
- Go SDK release
- Better error handling and retries
- Fixed PDF extraction for encrypted files
m
v1.8.0
- Rust SDK release
- On-premise deployment support
- Air-gapped installation mode
- SOC 2 Type II certification
m
v1.7.0
- Docker and Kubernetes deployment guides
- High availability clustering
- Helm chart for easy K8s deployment
m
v1.6.0
- Model fine-tuning support
- Custom extractor configuration
- Training data validation tools
m
v1.5.0
- Initial public release
- Python SDK with full API support
- CLI for local classification
- 12 file type extractors