I really enjoy having an opencode go subscription just so I can ask some less common models questions too. Sure DeepSeek. But MiMo, Kimi, MiniMax, Qwen... (Ok half those are not so unusual either.)

Agents cross comparing notes often surfaces some good improvements, finds interesting drifts. Ask them to reinterpret the prompt as they see it, have them describe the problem, then their findings, and run new rounds based on different models trying different prompts. Trying to swap and exchange ideas and vectors across agents.