Hermes Agent + DeepSeek V4 Pro
A CLI agent plus open/open-weight coding model stack for skills-based terminal experimentation.
ARscore 31.41M context$0.44/$0.87fully open
| Agent | Hermes Agent |
|---|---|
| Model | DeepSeek V4 Pro |
| Agent score | 39 |
| Model SWE-bench | 80.6% |
| Combined score | 31.4 |
| Best fit | skills-based terminal experimentation |
Related Hermes Agent stacks
Same agent, different model choices.
Hermes Agent + DeepSeek V4 Flash
ARscore 30.8, 1M context, $0.14/$0.28
view stack
Hermes Agent + Llama 4 Maverick
ARscore 24.2, 1M context, $0.8/$2.2
view stack
Related DeepSeek V4 Pro stacks
Same model, different agent wrappers.