Kimi K3
moonshotai/kimi-k3Moonshot AI~moonshotai/kimi-k-latestoperationalMoonshot AI Kimi K3, a frontier open-weight model with strong agentic and coding ability. 1M context.
Official pricing
USD per million tokens, exactly what Moonshot AI publishes. ElevenRouter adds no markup on tokens.
Example: 1,000 input + 500 output tokens ≈ $0.01050. Zero-completion responses are never charged.
Specs
- Context
- 1M
- Max output
- 64K
- Modalities
- text+image->text
- Tokenizer
- Kimi
- Released
- 2026-09-20
- Family
- kimi-k
Availability on ElevenRouter
Measured by the routing engine; up when at least one credential can serve the model.
Observed performance
Real requests over the last 7 days (1 samples).
- Time to first token p50
- —
- p95
- —
- Output tokens/s p50
- —
- p95
- —
Percentiles appear after five successful requests.
Usage on ElevenRouter
Tokens per day across all customers (anonymous, aggregate).
Use it
Any OpenAI or Anthropic SDK works by changing the base URL. Supported parameters: max_tokens, temperature, top_p, top_k, frequency_penalty, presence_penalty, repetition_penalty, seed, stop, tools, tool_choice, response_format, reasoning, include_reasoning.
curl https://elevenrouter.com/api/v1/chat/completions \
-H "Authorization: Bearer $ELEVENROUTER_API_KEY" \
-H "Content-Type: application/json" \
-d '{ "model": "moonshotai/kimi-k3", "messages": [{ "role": "user", "content": "Hello" }] }'Related models
- Claude Sonnet 4.6 · Anthropic$3.00 / $15.00
- GPT-5.6 Terra · OpenAI$2.00 / $12.00
- GPT-5.6 Sol · OpenAI$4.00 / $20.00
- Claude Sonnet 5 · Anthropic$2.00 / $10.00
- Claude Opus 4.6 · Anthropic$5.00 / $25.00
- Claude Opus 4.7 · Anthropic$5.00 / $25.00
Frequently asked questions
How much does Kimi K3 cost on ElevenRouter?
Kimi K3 is billed at Moonshot AI's official list price: $3.00 per million input tokens and $15.00 per million output tokens, with cached input at $0.300. A request with 1,000 input and 500 output tokens costs about $0.01050.
What is the context length of Kimi K3?
Kimi K3 accepts up to 1M tokens of context and can produce up to 64K output tokens.
Does Kimi K3 support tool calling and structured outputs?
Yes, tool calling is supported and JSON mode is available; add the response-healing plugin for schema validation. Supported parameters: max_tokens, temperature, top_p, top_k, frequency_penalty, presence_penalty, repetition_penalty, seed, stop, tools, tool_choice, response_format, reasoning, include_reasoning.
How do I call Kimi K3?
Send an OpenAI-compatible chat completion to https://elevenrouter.com/api/v1/chat/completions with "model": "moonshotai/kimi-k3" and your ElevenRouter key, or use the Anthropic Messages endpoint. Any OpenAI or Anthropic SDK works by changing the base URL.
Is Kimi K3 available right now?
Availability is measured continuously by the routing engine. Current status: operational, 100.00% uptime over the last 30 days. When one credential fails, requests fail over to another automatically.