Kimi (Moonshot AI) Commands & Prompt Reference

22 commands & parameters · 2 categories · Source: Kimi API Platform Docs

Moonshot's Kimi API is OpenAI-compatible on the surface, but its parameter set has its own character: max_completion_tokens stretches into the million-token range on Kimi K3, reasoning effort is always on and tunable between low, high, and max, and prompt_cache_key exists specifically so coding agents can keep cache hits high across a resumed session. This page lists the chat completion parameters and the model and mode tokens from the official platform documentation.

Partial Mode is the feature most worth memorizing: append an assistant message with partial set to true and your prefix in content, and the model is forced to open its reply with exactly that text - the documented way to pin a format or a first line.

All Commands

Chat API Parameters (16)

Command / SyntaxWhat it doesNotes
modelModel ID to call; kimi-k3 is the defaultSet to a model ID from the Models & Modes table below.
messagesThe conversation message list; supports dynamic tool messagesA {"role": "system", "tools": [...]} message can be inserted at any position.
max_completion_tokensMaximum tokens to generateKimi K3 defaults to 131072 and can be set up to 1048576. Replaces the deprecated max_tokens.
max_tokensDeprecated token limit fieldUse max_completion_tokens instead.
response_formatControls the output format; {"type": "json_object"} enables JSON modeDefault is {"type": "text"}. In JSON mode you must also guide the model to output JSON in the prompt.
stopStop words that halt generation on a full matchUp to 5 strings, each at most 32 bytes; matched words are not output.
streamReturn the response as a streamBoolean, default false.
stream_optionsOptions for streaming responsesObject with child attributes for stream behavior.
toolsA list of tools the model may callStandard function-tool definitions.
tool_choiceControls whether and how the model calls toolsauto (default), none, required, or an object forcing a specific function.
logprobsReturn log probabilities of the output tokensBoolean, default false; values arrive in the logprobs field of the response.
top_logprobsNumber of most likely tokens to return per positionInteger 0-20; requires logprobs: true.
predictionPredicted Output configuration to speed up largely-known responsesUseful when regenerating a file with only minor changes.
prompt_cache_keyCache key to raise cache hit rates across similar requestsFor coding agents, typically a session or task ID; reuse it when a session resumes.
safety_identifierStable per-user identifier for abuse detectionHash the username or email rather than sending it raw.
reasoning_effortReasoning effort level for Kimi K3 thinkingValues: low, high, max (default max). K3 always thinks with Preserved Thinking.

Models & Mode Tokens (6)

Command / SyntaxWhat it doesNotes
kimi-k3Flagship model ID with always-on thinkingDefault model; set as the model field value.
kimi-k2.7-codeCoding-focused model IDSet as the model field value.
kimi-k2.7-code-highspeedFaster variant of the coding modelSet as the model field value.
kimi-k2.6Previous-generation general model IDSet as the model field value.
kimi-k2.5Older general model IDSet as the model field value.
partial: truePartial Mode: force the reply to continue from a given prefixAppend a message with role: assistant, partial: true, and the prefix text in content at the end of messages.

Frequently Asked Questions

Which model ID should I default to?

kimi-k3 is the platform default and flagship, with always-on Preserved Thinking. The k2.7-code variants target coding workloads, with a highspeed option.

How do I force the reply to start with specific text?

Use Partial Mode: end the messages list with {role: assistant, partial: true, content: your-prefix}. The model continues from that prefix.

Does Kimi support JSON mode?

Yes - set response_format to {"type": "json_object"} and also instruct the model in the prompt to output JSON.

Every entry on this page is a literal, typeable token taken from the official documentation linked above. Command sets in AI tools change quickly - the in-product command menu is always authoritative for your installed version.