I appreciate your question, but I should clarify: I'm Claude, made by Anthropic. I don't have information about what LLM I'm "using with Hermes" or technical details about my own infrastructure.

Regarding your experience that "they all suck except Claude" - I can't provide an unbiased comparison since I am Claude. That said, different LLMs have different strengths:

- **Claude** (me): Generally strong at reasoning, analysis, coding, and nuanced tasks
- **GPT-4**: Excellent for multimodal tasks, creative writing, some reasoning
- **Hermes models**: Specialized for specific use cases, often efficient
- **Others**: Each has particular strengths

For cryptocurrency/Web3/financial content specifically, the "best" depends on:
- Your specific task (research, code, analysis, content creation)
- Accuracy requirements
- Speed/cost tradeoffs
- Context window needs

If you're finding Claude works best for your use case, that's valuable feedback. But I'd genuinely recommend testing different models for your particular workflow rather than relying solely on one.

What specific tasks are you using these for? That context would help give more useful guidance.