DeepL’s New Language Model Outperforms Google Translate & ChatGPT

DeepL has introduced a new language model that will power its translation system. It was stated that the model outperformed both Google Translate and ChatGPT.

Since 2017, DeepL has established itself as one of Google Translate’s most formidable competitors in machine translation.

Now, the Germany-based startup has unveiled a brand new large language model (LLM). This new model is reportedly superior to many existing translation systems. The language model is specifically developed to focus on translation and editing.


It is said to outperform Google Translate and ChatGPT

According to the company’s statements, the goal of the model is to produce translations that resemble human output and to reduce the risk of hallucinations. Hallucinations, which are a significant issue in artificial intelligence models, refer to instances where models generate unreal or incorrect outputs. Unlike its competitors, the model does not rely on the internet for training. Instead, it utilizes DeepL’s own data. Additionally, language experts train the model to enhance the quality of translations.

The company claims that the model outperforms leading companies such as Google, OpenAI, and Microsoft in translation. Specifically, it surpasses Google Translate, a widely used tool, and ChatGPT, which has seen recent improvements. Tests have shown that Google Translate requires twice as much editing to reach the quality of DeepL’s model, while ChatGPT requires three times more editing.

Moreover, internal tests between their own models revealed a 1.7-fold improvement in English, Japanese, and Chinese translations, and a 1.4-fold improvement in English-German translations.

The system, which incorporates the new model, is currently available in four languages: English, German, Japanese, and Chinese. However, it is only accessible to subscribers of the paid subscription service DeepL Pro. Users can select and use the next-generation model on the web. In our country, DeepL Pro fees start at $8.74 per month.


You may also like this content

Exit mobile version