Model Parameter Reference

Parameter Comparison

The K3 series column applies to kimi-k3, kimi-k3.1-pro, and kimi-k3.1-pro-highspeed. kimi-k2.7-code-highspeed has the same parameter constraints as kimi-k2.7-code.

Parameter	K3 series	`kimi-k2.7-code`	`kimi-k2.6`
Context window	1M tokens	256K tokens	256K tokens
`thinking`	Not supported	Optional. Thinking and Preserved Thinking are always on.	Optional. Thinking is on by default; reasoning from previous turns is not preserved by default.
`reasoning_effort`	`"low"` / `"high"` / `"max"` (default `"max"`)	Not supported	Not supported
`tool_choice`	`"auto"` / `"none"` / `"required"`	Passing `"required"` returns an error.	Passing `"required"` returns an error.
`temperature`	Fixed at `1.0`	Fixed at `1.0`	Fixed at `1.0` with thinking on and `0.6` with thinking off
`top_p`	Fixed at `0.95`	Fixed at `0.95`	Fixed at `0.95`
`n`	Fixed at `1`	Fixed at `1`	Fixed at `1`
`presence_penalty` / `frequency_penalty`	Fixed at `0`	Fixed at `0`	Fixed at `0`

“Fixed” means the parameter cannot be modified: passing any other value returns an error, so do not pass it explicitly. When temperature is close to 0, n can only be 1. Otherwise, the API returns invalid_request_error.

`thinking`

To turn off thinking for kimi-k2.6, set thinking.type to "disabled" in the request.

curl https://api.moonshot.ai/v1/chat/completions \
  --header "Authorization: Bearer $MOONSHOT_API_KEY" \
  --header "Content-Type: application/json" \
  --data '{
    "model": "kimi-k2.6",
    "messages": [{"role": "user", "content": "What is 1 + 1?"}],
    "thinking": {"type": "disabled"}
  }'

With thinking off, the response does not include reasoning_content. For supported values of thinking.type and thinking.keep, see Chat Completions.

`reasoning_effort`

K3 always reasons with Preserved Thinking enabled. Configure its reasoning effort with the top-level reasoning_effort request field, which supports "low", "high", and "max" (default "max"). See Reasoning Effort.

Switching levels invalidates prefix-cache hits. Decide on the effort level before the conversation starts and avoid switching it mid-session.

FAQ

How do I switch from kimi-k2.6 or kimi-k2.7-code to kimi-k3? Set model to kimi-k3 and remove thinking from the request. Use the top-level reasoning_effort field to set the reasoning effort. In multi-turn conversations and tool calls, pass the complete assistant message returned by the API back in messages, including reasoning_content. My code uses OpenAI’s reasoning_effort — do I need to change it for kimi-k3? No. K3 supports top-level reasoning_effort with "low", "high", and "max" as accepted values and "max" as the default.

Using the API

API Reference

Tools

Files

Batch API

Admin API

Parameter Comparison

`thinking`

`reasoning_effort`

FAQ