I was testing Kimi K3 shared endpoint on Modal.com using Aether OpenAPI-compatible endpoint, but couldn't find the option to toggle Reasoning/Thinking mode.
Meanwhile, when using curl to test it, there is reasoning option in the JSON payload:
"temperature": 0.3,
"max_tokens": 2048,
"top_p": 0.95,
"stream": false,
"reasoning": {"enabled": true},
However, when not using thinking mode, it use "reasoning_effort" instead of "reasoning" (may be the default value of "reasoning" is already false):
"temperature": 0.3,
"max_tokens": 2048,
"top_p": 0.95,
"stream": false,
"reasoning_effort": "none",
I was testing Kimi K3 shared endpoint on Modal.com using Aether OpenAPI-compatible endpoint, but couldn't find the option to toggle Reasoning/Thinking mode.
Meanwhile, when using
curlto test it, there isreasoningoption in the JSON payload:However, when not using thinking mode, it use "reasoning_effort" instead of "reasoning" (may be the default value of "reasoning" is already false):