base_model = "moonshotai/kimi-k2.5"
reasoning_options = []

[interleaved]
field = "reasoning_content"

[cost]
input = 0.6
output = 3
cache_read = 0.1