The new version of Deepseek at the forefront of non-reasoning AI models

DeepSeek has taken an important step in artificial intelligence development with the launch of DeepSeek V3-0324, which has become the highest-performing open-access AI model in the category of non-reasoning models, according to the analysis by the independent platform Artificial Analysis.

This milestone is significant because, for the first time, an open-source model leads this category, surpassing even models from giants such as OpenAI (ChatGPT), Google (Gemini 2.0 Pro), Anthropic (Claude 3.7 Sonnet), and Meta (Llama 3.3 70B).

DeepSeek V3-0324 marks the first time an open weights model has been the leading non-reasoning model. pic.twitter.com/TXwowHQA9x

— Artificial Analysis (@ArtificialAnlys) March 25, 2025

The success of DeepSeek V3-0324 in this ranking is due to several factors. Firstly, its specialization in speed and low latency differentiates it from deep reasoning models, as it is designed to provide immediate responses without needing to “think.” This feature makes it ideal for applications where speed is crucial, such as real-time chatbots and automated assistants. Moreover, unlike other high-performing models that are often closed, DeepSeek has succeeded in competing directly with giants like OpenAI and Anthropic with an open-access model, allowing the research community and independent companies to leverage its technology without restrictions.

🚀 DeepSeek-V3-0324 is out now!

🔹 Major boost in reasoning performance
🔹 Stronger front-end development skills
🔹 Smarter tool-use capabilities

✅ For non-complex reasoning tasks, we recommend using V3 — just turn off “DeepThink”
🔌 API usage remains unchanged
📜 Models are… pic.twitter.com/QVuPwCODne

— DeepSeek (@deepseek_ai) March 25, 2025

This launch not only reinforces DeepSeek’s leadership in open-access models but also suggests that the upcoming model DeepSeek R2 could represent another significant leap in the field of artificial intelligence. With this advancement, DeepSeek demonstrates that open-source AI models can match or even surpass proprietary ones in certain categories, marking a turning point in the AI ecosystem.

2025: The Year of DeepSeek

After its viral popularization, Deepseek was positioned ias one of the most noteworthy launches of the year in the artificial intelligence sector, despite continuous calls for caution by various regulatory bodies. Consequently, DeepSeek was blocked in Italy due to concerns about its security.

• To know more: 10 questions everyone asks about DeepSeek

In light of such news, the initial enthusiasm was short-lived.  DeepSeek hit a maximum of 54.67 million visits in January, but then decreased to 22 million. As the weeks went by, the trend was quite similar.

We will see how these figures evolve with the arrival of this new model.

Image: Flux Schnell