v10.8.0

• Support switch whipser model
• Support continue download whisper model if lost connection
• Remove embedding model
• Support macOS 12.4 or later

v10.7.0

• New logo
• Support Polish
• Improve settings view
• Improve translate speed, close think of LLM
• Build with macOS 26

v10.6.0

• Support export audio

v10.5.0

• Improve translation accuracy and complementation. For example, gemma3:1b model's complementation is 6x

v10.4.0

• Fix bug: transcribe audio failed in mac which has only 8G memory
• Use Gemma3:4b as default LLM model, suitable for the little memory size mac

v10.3.0

• Improve audio transcribe speed with 3x
• Improve translate speed with 5x

v10.2.0

• Tuning model download
• Fix translation with LLM timeout bug

v10.1.0

• Memorize video play time
• Support export source, translation, dual language subtitles if click export srt button

v10.0.0

• Support reasoning model. For example: Deepseek-R1-xxx, QwQ etc.
• Tuning subtitle translate accuracy
• Speed subtitle translate
• Support OpenAI model
• Support select source language
• Support download models from internet
• Support statistics translated success ratio

v9.11.0

• Memorize last selected language
• Memorize last selected model
• Fix bug: exported srt file do not include translation after import subtitle file