DeepSeek: A Multi-modal Flash Beats Claude Opus 4.8 on Agent Tasks From DeepSeek
DeepSeek introduces the V4-Flash-Vision-Exp: a multi-modal model that matches Anthropic’s Claude Opus 4.8 on agentic tasks (meaning it solves problems sequentially on its own) and can handle up to 600 images per request.
The V4-Flash-Vision-Exp model can process images, screenshots, and graphs. It achieves comparable performance to Anthropic’s Claude Opus 4.8 on agentic prompts and can process up to 600 images at once for the same price as the default V4-Flash model (which only supports text).
A year ago, these types of multi-modal models were expensive and out of reach for most people. As DeepSeek continues to push down the cost of these models, they’ll be putting even more pressure on companies like OpenAI and Anthropic.
Source: the-decoder.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Founder & CEO, Dad4Kids
Articles · Latest articles