Short answer
Kimi K3.1 Pro is Moonshot AI's new flagship, released on October 24, 2026. On TeamoRouter its model ID is kimi-k3.1-pro, with a 1,000,000-token context window and text, image and video input. It costs $2 per million input tokens and $10 per million output tokens, with cache reads at $0.20, the same as Moonshot's list price and cheaper than Kimi K3 ($3 / $15). To run it in Claude Code, point ANTHROPIC_BASE_URL at TeamoRouter and set ANTHROPIC_MODEL=kimi-k3.1-pro. You don't need a Moonshot account, and the same key works for Claude, GPT and Gemini.
What is new in K3.1 Pro
From Moonshot's K3.1 Pro guide:
- First model of the K3.1 series. It is Kimi K3 with much more post-training, including large-scale reinforcement learning across hundreds of domains.
- Open weights. Moonshot calls it a 3T-class open-weight model and says the full weights will be released by November 4.
- Where it improved most: coding, 3D creation (parametric modeling of parts, buildings and furniture), and financial research and modeling.
- Reasoning is always on. You can't turn thinking off; on Moonshot's API you control depth with
reasoning_effort(low,high,max, defaultmax). - A faster variant.
kimi-k3.1-pro-highspeedruns about 2 to 3 times faster at twice the price. It is not on TeamoRouter yet.
Moonshot recommends K3.1 Pro for coding agents such as Claude Code, Codex and Kimi Code. Its published benchmark figures were still placeholders on launch day, so this article doesn't quote any.
Kimi K3.1 Pro API pricing per million tokens
Moonshot's list prices, in USD:
| Model | Input | Output | Cache read | Cache write (5 min) |
|---|---|---|---|---|
kimi-k3.1-pro |
$2.00 | $10.00 | $0.20 | $2.00 |
kimi-k3.1-pro-highspeed |
$4.00 | $20.00 | $0.20 | $4.00 |
kimi-k3 |
$3.00 | $15.00 | $0.30 | $3.00 |
On TeamoRouter, kimi-k3.1-pro is billed at the same $2 / $10, with cache reads at $0.20. Pricing is flat across the whole 1M context, with no long-context surcharge. Live prices are on the model page.
What a session costs: a mid-sized Claude Code task that reads about 200k input tokens and writes 20k output tokens comes to about $0.40 + $0.20 = $0.60 on K3.1 Pro, before cache savings. For comparison, the same task on Claude Sonnet 5.5 through TeamoRouter is about $0.17, so pick K3.1 Pro for what it does well rather than to save money.
Set up Kimi K3.1 Pro in Claude Code
You need a TeamoRouter API key, which you can create in the console. If Claude Code isn't installed yet, follow the installation guide first, then run claude update so your version is current.
macOS and Linux
Add to ~/.zshrc (or ~/.bashrc):
export ANTHROPIC_BASE_URL=https://api.teamorouter.com
export ANTHROPIC_AUTH_TOKEN="your TeamoRouter key"
export ANTHROPIC_API_KEY=""
export ANTHROPIC_MODEL=kimi-k3.1-pro
export ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-haiku-4-5
Then source ~/.zshrc and run claude. Don't add /v1 to the base URL; Claude Code adds the path itself. ANTHROPIC_DEFAULT_HAIKU_MODEL keeps Claude Code's small background tasks on a cheap model.
Windows (PowerShell)
setx ANTHROPIC_BASE_URL "https://api.teamorouter.com"
setx ANTHROPIC_AUTH_TOKEN "your TeamoRouter key"
setx ANTHROPIC_API_KEY ""
setx ANTHROPIC_MODEL "kimi-k3.1-pro"
setx ANTHROPIC_DEFAULT_HAIKU_MODEL "claude-haiku-4-5"
Close and reopen the terminal after setx.
Check it works
claude -p "Which model are you? Reply in one line."
Inside a session, /status shows the model in use, and /model claude-sonnet-5-5 switches to Claude for a task without changing your setup.
Call the API directly
TeamoRouter accepts both the OpenAI and the Anthropic format for this model.
OpenAI SDK:
from openai import OpenAI
client = OpenAI(
api_key="your TeamoRouter key",
base_url="https://api.teamorouter.com/v1",
)
resp = client.chat.completions.create(
model="kimi-k3.1-pro",
messages=[{"role": "user", "content": "Write a Python function that parses ISO 8601 durations."}],
)
print(resp.choices[0].message.content)
Anthropic SDK:
import anthropic
client = anthropic.Anthropic(
api_key="your TeamoRouter key",
base_url="https://api.teamorouter.com",
)
msg = client.messages.create(
model="kimi-k3.1-pro",
max_tokens=2048,
messages=[{"role": "user", "content": "Explain this stack trace and suggest a fix."}],
)
print(msg.content)
Because K3.1 Pro always reasons, leave enough room in max_tokens for the thinking plus the answer.
Paying for usage
TeamoRouter bills per token with no subscription. Top up from $5 by card, or with USDT, USDC or ETH from any exchange app: click Pay with crypto in the top-up dialog. Step by step: Pay for Claude and GPT APIs with crypto (USDT). One balance covers Kimi, Claude, GPT and Gemini.
FAQ
What is the Kimi K3.1 Pro model ID? kimi-k3.1-pro, the same on TeamoRouter as on Moonshot's own API.
How is K3.1 Pro different from Kimi K3? It is K3 with much more post-training, focused on coding, agent work, 3D modeling and finance. It is also cheaper at list price: $2 / $10 against $3 / $15.
Is Kimi K3.1 Pro open source? Moonshot describes it as an open-weight model and says the full weights will be published by November 4, 2026. Until then you use it through an API.
Can I turn off thinking? No. K3.1 Pro always reasons. On Moonshot's API, reasoning_effort: "low" shortens the reasoning when latency matters.
Can I send images? Yes. Moonshot's docs say image input must be base64 or an uploaded file ID; public image URLs are not accepted.
Do I need a Moonshot (Kimi) account? No. With a TeamoRouter key you call kimi-k3.1-pro directly, and the same key and balance work for other models.
Can I pay for Kimi K3.1 Pro with USDT? Yes. Click Pay with crypto in the TeamoRouter top-up dialog and send USDT, USDC or ETH from Binance, OKX or any wallet, from $5. See how to pay with USDT.