DeepSeek has made a name for themselves in the artificial intelligence arena in the past year, when they successfully came up with a powerful and comparable artificial intelligence model, but at a much lower development cost. Now, a year after DeepSeek v3.2 became the hot topic of many parties and the industry, DeepSeek today officially showcased two versions under DeepSeek v4, namely DeepSeek v4 Flash and v4 Pro.
Through this new model, DeepSeek offers a context of up to 1 million tokens, allowing it to remain consistent in generation, and is suitable for the development of large code or documents. This new model is offered in open source form, allowing anyone interested to download and use it in development.
This model also emphasizes the reasoning and task capabilities of AI agents compared to other models on the market. For now, it still only supports text. Both v4 models are also more efficient than the previous DeepSeek v3.2 model, in line with the development of technology and technical design.
For those of you who are interested, this new model can be tested via the DeepSeek website under Expert or Instant mode. Updates to the API have also started being offered starting today.
