1.5.14

Updated

• Fix a bug failing to load a big-sized model when loaded from the local cache.

1.5.13

Updated

• Added Lite tier support
• Improved Gemma 4 support for LLM
• Fixed an issue preventing LLM models from running in some builds

1.5.12

Updated

• Add Lite tier in the enum
• Update LLM for support gemma4

Known Issues

• LLM initialization may fail in some distributed iOS builds, preventing chat or model execution.
• If affected, update to the latest release.

1.5.11

Updated

As the same as 1.5.10

1.5.10

Updated

• Update ZeticMLangeLLMModel API
  • Support selecting apType (.CPU / .GPU) together with quantType when target is .LLAMA_CPP
  • Add initOption to configure LLM initialization options
  • Move kvCacheCleanupPolicy from the initializer arguments into initOption
  • Support configuring context length with initOption.nCtx

TBD

• Update docs about the new ZeticMLangeLLMModel API

1.5.9

Updated

• Add cacheHandlingPolicy and ModelCacheManager.
• Update default modelMode in ZeticMLangeLLLMModel.

TBD

• Update docs about cache handling

1.5.8

Release version 1.5.8

1.5.7

Full Changelog: 1.5.6...1.5.7

1.5.6

Full Changelog: 1.5.5...1.5.6

1.5.5

Full Changelog: 1.5.4...1.5.5