What are the differences between Generative AI engines?

【Security】 

No generative AI engines reuses the data input by users for secondary purposes.

Here is the information about the data storage and processing servers.

• GPT-5.4 and GPT-5.4 mini:  The data is not stored anywhere (zero retention policy applied), but the data can be processed outside Japan (US)

• Yaraku:  The data is not stored anywhere, and the data is processed in Japan

• Claude: The data is not stored anywhere, but the data can be processed outside Japan (US Virginia and US Oregon). 

• Gemini 3.0 Flash: The data is not stored anywhere, but the data can be processed outside Japan (US). 

【Translation processing speed】 

•  GPT-5.4 mini is a lightweight model with fast processing speed and reduced latency

• Gemini 3.0 Flash has fast processing speed, but has limit on number of tokens to be handled at a time

• GPT-5.4 has normal processing speed, but still a bit slower than conventional machine translation engines

• Yaraku engine has a slower processing speed in comparison to Gemini and GPT engines to its in context translation processing

• Claude has the slowest processing speed