Kimi K3.1 Pro API
Moonshot AIText, 1M context, answers up to 32K tokens
Moonshot AI's most capable model: long coding sessions, research and financial modeling. 1M tokens of context, OpenAI's chat format, pay per token.
- Input$1.65
$2−18% - Output$8.25
$10−18%
- A question500 tokens in, 300 out$0.0033
$0.004 - A long chat10K in, 1K out$0.0248
$0.03 - A document100K in, 2K out$0.1815
$0.22 - A month of a busy bot50M in, 10M out$165.00
$200.00
Official: Moonshot AI's price for Kimi K3.1 Pro, cache miss, Kimi API pricing, checked on Nov 14, 2026.
API
Call Kimi K3.1 Pro in code.OpenAI's format: change the base URL and the key.
POST /v1/chat/completions
- model
kimi-k3.1-pro- messages
- The conversation: system, user and assistant turns, as in OpenAI's API. Text and images in user turns.
- max_tokens
- The longest answer, up to 32,000 tokens. 8,192 when not set.
- stream
- true for server-sent events as the answer is written.
- …
- temperature, tools, response_format and the rest go to the model as they are.
curl https://api.tikiapi.com/v1/chat/completions -H "Authorization: Bearer $TIKIAPI_KEY" -H "Content-Type: application/json" -d '{
"model": "kimi-k3.1-pro",
"messages": [
{
"role": "user",
"content": "Explain compound interest with a short example: $1,000 at 5% a year for 3 years."
}
],
"max_tokens": 1024
}'Overview
What Kimi K3.1 Pro is good at.Moonshot AI's flagship for long-horizon coding and knowledge work.
Long-horizon coding
Big tasks carried through many steps.
Knowledge work
Research, writing and financial models.
1M-token context
Long documents and code in one call.
Pay per token
Input and output priced per million. No plan.
Use cases
Coding
Features and refactors end to end.
Research
Reports from many sources.
Finance
Models, forecasts and analysis.
Apps and automation
Kimi inside your product, bot or pipeline.
FAQ
Kimi K3.1 Pro API questions
Something else? Read the docs
How much does the Kimi K3.1 Pro API cost?
$1.65 per million input tokens and $8.25 per million output tokens. A question of 500 tokens with a 300-token answer costs $0.0033. No subscription, nothing upfront.
Does it work with the OpenAI SDK?
Yes. Point the SDK at https://api.tikiapi.com/v1, use your TikiAPI key and the model id kimi-k3.1-pro. The request and the answer have OpenAI's chat completions format, streaming included.
How long can an answer be?
Up to 32,000 tokens, set with max_tokens. Without it the answer is capped at 8,192 tokens.
What if a call fails?
It costs nothing. You pay for the tokens of an answer you get, never for an error.
Other text models.
Anthropic · Text
Claude Opus 5.5
Anthropic's model for long-running agentic coding and knowledge work.
$2 in · $10 outper 1M tokens · 1M context
Anthropic · Text
Claude Sonnet 5.5
Anthropic's best mix of speed and intelligence.
$1 in · $5 outper 1M tokens · 1M context
Anthropic · Text
Claude Haiku 5.5
Anthropic's fast model for high-volume work.
$0.05 in · $0.25 outper 1M tokens · 1M context
Anthropic · Text
Claude Fable 5.2
Anthropic's model for demanding reasoning and long-horizon agentic work.
$4 in · $20 outper 1M tokens · 1M context
OpenAI · Text
GPT-6.1 Sol
OpenAI's near-Astra model for complex work at a lower cost.
$1 in · $5 outper 1M tokens · 1.05M context
OpenAI · Text
GPT-6 Astra
OpenAI's most capable model for the most demanding work.
$5 in · $25 outper 1M tokens · 1.05M context
Google · Text
Gemini 3.8 Flash
Google's most capable Flash model, built for agents and long software tasks.
$0.40 in · $2 outper 1M tokens · 1M context
Google · Text
Gemini 4 Argon
Google's efficient Gemini 4 reasoning model for code, agents, math and science.
$1 in · $5 outper 1M tokens · 1M context
