DeepSeek Releases Experimental Multimodal AI Model
DeepSeek launched the DeepSeek-V4-Flash-Vision-Exp model, adding visual analysis and action capabilities to its existing text-only flagship series.
DeepSeek released an experimental multimodal model, DeepSeek-V4-Flash-Vision-Exp, on Friday. The new tool integrates visual prompt analysis and action capabilities into the company's existing text-only V4 Flash flagship series.
The Hangzhou-based company claims the model's multimodal agentic performance is close to that of the Opus 4.8 model developed by Anthropic PBC. This release follows a trend of Chinese AI firms providing high-performance models at lower costs than their United States competitors.
Separately, Anthropic PBC is preparing for a potential initial public offering as early as the end of August, seeking to match or exceed the record size of the SpaceX IPO. While Anthropic raised $65 billion in May at a $965 billion valuation, the company reported a net loss of nearly $42 billion in 2025.