Use a vision model helper to improve Deepseek inside Hermes
and Pro do not directly handle images, so results can fall short when a task depends on a screen, picture, or . In Hermes, a separate can be called as a sub-agent to cover that gap.
The first turns the visual problem into a clear text description, then uses that description to produce the answer. This can create a stronger without forcing the user to explain every visual detail by hand.
Firsthand testing in Hermes and Kilo found much better results after prompting the agent to work this way. The sub-agent handoff is still not as smooth as top , so the setup may need clear instructions.
Key points
- and Pro lack built-in image understanding.
- Hermes can use a as a sub-agent to describe images or screens.
- The main model can then answer from the visual description instead of guessing.
- The approach worked better in Hermes and Kilo when the agent was prompted clearly.
- Sub-agent delegation may still be uneven, so give explicit step-by-step instructions.