Deepseek releases experimental Flash vision model that rivals Opus 4.8 on agent benchmarks
decoder ·
Deepseek has released V4-Flash-Vision-Exp, an experimental multimodal model that adds image understanding to its text-focused V4-Flash base. According to Deepseek's own benchmarks, the model nearly matches Opus 4.8 on agent tasks while preserving the base model's text performance. Aimed at visual agent workflows, it can describe images, extract text from screenshots, and analyze diagrams, and works with various agent frameworks. It supports JPEG, PNG, GIF, and WebP, and is compatible with OpenAI's Chat Completions and Responses APIs and Anthropic's Messages endpoint. Deepseek also updated its Harness framework to version 0.1.1 to support the model. Images can be sent via Base64, public URLs up to 32 MiB, or a free Files API with a 64 MiB limit; each image costs at most 384 tokens, with up to 600 images per request.
decoder