Noesion Radarnoesion.ai
Release2026-08-21

Deepseek released V4-Flash-Vision-Exp, an experimental version of its Flash model that can read images as well as text. On Deepseek's own tests of agent tasks, it comes close to and sometimes beats Anthropic's Opus 4.8, though those figures are self-reported. Pricing matches the existing Flash rates, and images can only be sent in user messages.

What changed

V4-Flash handled text only, with no image input.

What it unlocks

Sending screenshots, diagrams and photos to a Deepseek model inside agent workflows, and re-using one uploaded file across many requests via the free Files API.

  • max 384 tokens per image
  • up to 600 images per request
  • Files API limit 64 MiB
  • image URLs up to 32 MiB

What you need to act on it

  • API access to Deepseek

Send this to someone who needs it

Shares the story and its sources. Nothing about you.

What does this mean for your job?

This is the story as everyone gets it. Once a week we send you the version written for your role — what changed, why it matters for the work you actually do, and one thing to try. Free while we tune it.