"DeepSeek's V4 model includes two versions: DeepSeek-V4-Pro, which has 1.6 trillion parameters, and DeepSeek-V4-Flash, a smaller variant. The Pro version's context length is 1 million tokens, equating to approximately 750,000 words."
"The release of V4 follows the significant market impact of DeepSeek's previous models, which reset conversations around AI and demonstrated that frontier-level performance could be achieved on a limited budget."
"DeepSeek claims its new model outperforms all other open-source models in agentic coding and reasoning, raising questions about the competitive advantages of leading U.S. labs like OpenAI and Anthropic."
DeepSeek launched its V4 large language model, which includes two versions: DeepSeek-V4-Pro and DeepSeek-V4-Flash. The Pro version features 1.6 trillion parameters and a context length of 1 million tokens. The release follows the significant impact of its previous models on the AI market. While V4 may not affect markets as dramatically, it has influenced the shares of Chinese companies, particularly Semiconductor Manufacturing International Corporation. DeepSeek claims its model excels in agentic coding and reasoning, raising questions about the competitive landscape in AI development.
Read at Fortune
Unable to calculate read time
Collection
[
|
...
]