The latest release of DeepL's next-generation language model has achieved significant breakthroughs in translation accuracy, and is said to have surpassed Google Translate and GPT-4. After more than seven years of data training, this model has been deeply optimized for the fields of translation and proofreading. It currently supports translation between Japanese, English, German and Simplified Chinese, and plans to gradually expand to more languages. Its superiority is reflected in the blind test results, which show significant improvements in multi-language translation and significantly reduce the workload of post-editing.
DeepL recently announced the launch of its new generation language model, saying that the model has surpassed Google Translate and GPT-4 in translation accuracy. The new generation language model uses more than seven years of unique data for training, focuses on optimization in the field of translation and proofreading, supports translation into Japanese, English, German and Simplified Chinese, and plans to expand to support more languages in the future.
According to DeepL's blind test results, the new model performs better than the traditional model in terms of translation quality. Tests show that the performance of the new model has improved by 1.7 times in "Japanese-English translation" and "Simplified Chinese-English translation", and by 1.4 times in "German-English translation".
In addition, in comparison with Google Translate, GPT-4 and Microsoft Translate, the frequency of the new model being rated as "better translation quality" was respectively 1.3 times higher than that of Google Translate, 1.7 times higher than that of GPT-4, and 1.7 times higher than that of Microsoft Translate. 2.3 times higher. To achieve the same translation quality as the new model, Google Translate requires twice as much editing work, while GPT-4 requires three times as much editing work.
Currently, the new generation language model is online in the DeepL Pro plan, and users can switch to the new model simply by pressing a button. The monthly fee for the DeepL Pro plan starts from 1,150 yen (tax included) and is divided into three packages: Starter, Advanced and Ultimate.
Highlights:
The new generation language model released by DeepL surpasses Google Translate and GPT-4 in translation accuracy.
After blind testing, the new model performed better in Japanese-English translation, Simplified Chinese-English translation, and German-English translation, and the editing workload was significantly reduced.
? The DeepL Pro plan already supports new models, users can switch with a button, and the cost starts from 1,150 yen/month.
All in all, the launch of DeepL's new generation language model marks a new milestone in machine translation technology, providing users with more efficient and accurate translation services. Its application prospects in professional fields are worth looking forward to, and it also indicates the future development direction of machine translation technology.