Moonshot AI (Kimi) / moonshot/kimi-k3

Kimi K3 - access through LLMTR

Moonshot AI Kimi K3 is the current flagship Kimi model for long-context analysis, multi-step agent workflows and visual understanding. It supports a 1,048,576-token context window, image input (base64 data URLs), function calling, JSON output and automatic prompt caching. Thinking is on by default and can be turned off with reasoning:false in the request body or the :fast model suffix. Sampling parameters are fixed (temperature=1, top_p=0.95); tool_choice accepts auto, none or required. Web search is not supported on this model. Pricing is $0.30/1M cached input, $3.00/1M cache-miss input and $15.00/1M output. Reasoning tokens are billed as output; keep max_tokens generous even for short answers.

Technical specifications

Canonical IDmoonshot/kimi-k3
ProviderMoonshot AI (Kimi)
Context window1,048,576 tokens
OperationsCHAT_COMPLETIONS
Modalitiestext, image

Pricing

A 6% platform margin applies to credit top-ups; model usage prices are not separately marked up.

OperationMetricUnitPrice
CHAT_COMPLETIONSINPUT_TEXTPER_1M_TOKENS$3.00
CHAT_COMPLETIONSCACHE_READPER_1M_TOKENS$0.300000
CHAT_COMPLETIONSOUTPUT_TEXTPER_1M_TOKENS$15.00

Example usage

With existing OpenAI SDK flows, change only the base URL and model identifier.

curl https://llmtr.com/v1/chat/completions \
  -H "Authorization: Bearer llmtr-your_key" \
  -H "Content-Type: application/json" \
  -d '{"model":"moonshot/kimi-k3","messages":[{"role":"user","content":"Hello"}]}'

Related models