DeepSeek V4 Flash Nears Opus 4.8 on Multimodal Tests
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- DeepSeek unveiled an experimental multimodal version of its V4 Flash model, claiming it nears the performance of Anthropic's Opus 4.8 on multimodal agentic tests.
- The experimental model can understand visual prompts, according to DeepSeek's announcement covered by Bloomberg.
Why it matters: DeepSeek's self-reported framing positions a Chinese-built model against a top-tier Western frontier benchmark on agentic multimodal tasks. The claim is unverified by independent testing and the model is explicitly experimental, so the practical gap to Opus 4.8 remains to be confirmed.
Ask SkimNews
