hi folks!
how does thinking mode and reasoning effort work for DeepSeek models in KeiRouter?
if i use CLAUDE_CODE_EFFORT_LEVEL=max with deepseek-v4-flash, will KeiRouter automatically enable thinking mode and send reasoning_effort=max to DeepSeek?
or do i need to use a separate -max model?
hi folks!
how does thinking mode and reasoning effort work for DeepSeek models in KeiRouter?
if i use CLAUDE_CODE_EFFORT_LEVEL=max with deepseek-v4-flash, will KeiRouter automatically enable thinking mode and send reasoning_effort=max to DeepSeek?
or do i need to use a separate -max model?