AI BriefingDeepSeek V4 Flash Vision Experimental Version Now Available on AI Gateway
Key point
An experimental vision-capable version of DeepSeek V4 Flash has been released on Vercel AI Gateway.
Details
Vercel has released the DeepSeek V4 Flash Vision Experimental Version on AI Gateway, which can process text and images simultaneously. This model can handle visual information alongside text prompts, such as reading text from screenshots or analyzing charts.
Existing tool use, reasoning, and caching features are supported as before. The model ID is deepseek/deepseek-v4-flash-vision-exp, and it supports images in JPEG, PNG, GIF, and WebP formats. Since the format is determined based on actual byte data rather than file extensions, it processes correctly even if the filename is incorrect.
The -exp included in the model ID indicates an experimental release, and it is recommended to configure a fallback model in production environments to prepare for behavioral changes. To integrate with coding agents (such as Claude Code, Codex, etc.), run the vercel ai-gateway coding-agents setup command and then select the model.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.