Dsh Vision Mix
haiziyao/dsh-vision-mix · model-adapters
Visual routing and image generation for DeepSeek Harness is done through a fixed Mix model.
★ 4GitHub stars
1Forks
2026-09-08Last updated
TypeScriptLanguage
MITLicense
Key features
- Many transit stations/models only return models id, and do not state whether the model will accept pictures.
- Check in turnwhether the model id is correct, whether the transit station accepts photo messages, whether the protocol is consistent with the transit station, and whether API Key is available.
- Vivation mix image generateGenerates new pictures based on hints.
- The user uploads the picture, displays the original message and then calls for visual model analysis.
- DSH will safely consider these self-defined models as pure text models.
Requirements
- 2-rc.1 Local tarball installation, configuration loading, Web cold start, HTTP access and offload validation has been completed in Windows x64 for one-time Profile.
- 3-alpha.2 The official CLI is blocked on this machine by its own fs-ext primary dependence, so both versions are conservatively marked unknown.
Install command
dsh plugin --profile web add github:haiziyao/dsh-vision-mix