What happened
Chinese AI company DeepSeek released V4-Flash-Vision-Exp, an experimental multimodal model that adds image understanding to its text capabilities. On DeepSeek's internal multimodal agent benchmarks, the model scores close to Opus 4.8.
Why it matters
DeepSeek-V4-Flash-Vision-Exp extends the base text model while preserving its reasoning and world knowledge performance. The vision variant is designed for agent-based applications—it can describe images, extract text from screenshots, analyze diagrams, and work with OpenAI's Chat Completions and Responses APIs as well as Anthropic's Messages endpoint.
What to watch
Developers can send images three ways—Base64 encoding, publicly accessible URLs (up to 32 MiB), or DeepSeek's free Files API (64 MiB limit). A single request supports up to 600 images; each image costs at most 384 tokens, and pricing follows V4-Flash rates. DeepSeek also released version 0.1.1 of its Harness framework, which supports the new model out of the box.
Summaries like this, in your inbox every morning.
DeepSeek-V4-Flash-Vision-Exp represents the company's push into multimodal agent workflows, extending its existing V4-Flash model with vision capabilities while maintaining the base model's text performance in reasoning and world knowledge. The model's near-parity with Opus 4.8 on DeepSeek's internal multimodal agent benchmarks suggests competitive positioning in the visual reasoning space. The compatibility with both OpenAI's and Anthropic's APIs—Chat Completions, Responses APIs, and the Messages endpoint—signals that DeepSeek is designed for integration into existing workflows rather than as a standalone alternative. The release of version 0.1.1 of DeepSeek's Harness framework with out-of-the-box support for the new model underscores the company's investment in developer tooling and ecosystem development.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Chunghwa Telecom president Rong-shy Lin said the shift from conversational AI to autonomous AI agents is resha…
Apple and Meta said they would jointly release a new open-weight AI model designed to run on high-end Macs, ai…
A writer and entrepreneur opened his daily to-do list and found 156 items, and concluded that an AI assistant…
A Zenn contributor sorted AI use into three modes — AI as an assistant, AI running alongside people, and AI th…
The walkthrough points the CLI at a local mock via the ANTHROPIC_BASE_URL environment variable
A non-engineer operator says one person ships four smartphone apps on Google Play, three on the App Store, wit…
