Kimi (Moonshot)

Vendor: Moonshot AI · Official API docs: platform.moonshot.ai

Kimi K3 is Moonshot AI's flagship text model — deep reasoning with a 1M-token context, and it accepts up to 8 images alongside the prompt. Use it for long-document analysis, image understanding and multi-step reasoning. Like every text model in the catalog it is driven with gen-ai ask rather than gen-ai generate.

Mode: text · Models: 3

Models

id	Name	Input type
kimi-k3	Kimi K3	i2t
kimi-k3.1-pro	Kimi K3.1 Pro	i2t
kimi-k3.1-pro-highspeed	Kimi K3.1 Pro Highspeed	i2t

CLI

# reasoning over a long prompt (thinking defaults to high)
gen-ai ask -m kimi-k3 -p "Summarize the key risks in this plan and rank them"

# image understanding, with a faster thinking budget
gen-ai ask -m kimi-k3 -p "What does this chart show?" -i ./chart.png --thinking low

MCP

{ "name": "picsart_generate",
  "arguments": {
    "model": "kimi-k3",
    "prompt": "Summarize the key risks in this plan and rank them",
    "thinking": "high"
  } }

Parameters

Full parameter surface for every model, sourced from gen-ai models info <id> --json. CLI flags show the primary short form; the canonical --kebab-case long form always works too.

kimi-k3 — Kimi K3

Try kimi-k3 in Playground ↗

Input type: i2t

Param	CLI flag	Type	Values
prompt	-p	text	required
imageUrls	-i	file	image (up to 8)
thinking	--thinking	enum	low · high (default high)

kimi-k3.1-pro — Kimi K3.1 Pro

Try kimi-k3.1-pro in Playground ↗

Input type: i2t

Param	CLI flag	Type	Values
prompt	-p	text	required
imageUrls	-i	file	image (up to 8)
thinking	--thinking	enum	low · high (default high)

kimi-k3.1-pro-highspeed — Kimi K3.1 Pro Highspeed

Try kimi-k3.1-pro-highspeed in Playground ↗

Input type: i2t

Param	CLI flag	Type	Values
prompt	-p	text	required
imageUrls	-i	file	image (up to 8)
thinking	--thinking	enum	low · high (default high)