MoonshotAI: Kimi K3.1 Pro API
Moonshot AI's flagship open-weight multimodal reasoning model with a 1M-token context, native image and video input, and strong coding, 3D, and finance performance.
Kimi K3.1 Pro is Moonshot AI's flagship open-weight multimodal reasoning model and the first release in the K3.1 series, launched on October 23, 2026. It is a post-training upgrade of Kimi K3, described by the company as a 3T-class model with a total parameter count reported to match K3's 2.8 trillion. It is offered through the Kimi website, Kimi Work, Kimi Code, and the API.
The model has native visual understanding: it accepts image and video input and offers a 1M-token context window for very long text and complex tasks. Compared with the previous generation, coding, 3D creation, and financial research improved notably. On Terminal-Bench 4.0 it scores 65.7%, close to Claude Opus 5.5 at 66.4% and above GPT-6 Astra at 58.2%; internal tests put financial research at 70.4 points, up from 58.8, and financial modeling at 57.6, up from 50.4. Individual-statement hallucination rate on Kimi Hallu Bench Hard dropped from 23.4% to 8.3%.
It fits long-horizon coding, 3D model and game design, financial statement analysis and investment research, financial modeling, and end-to-end agentic knowledge work. Multi-agent orchestration scales to hundreds of parallel subagents for autonomous task decomposition. The model is open-weight, with full weights planned for release before November 4, 2026, and its cyber-offense capability is deliberately capped as a safety measure.
Updated:
MODEL ID
kimi-k3.1-pro- Context
- 1.049M
- Max output
- 943.718K
- Recommended API
- messages
- Input
- text / image / video
- Output
- text
Highlights
- 1M-token context window
- Native image and video input
- Terminal-Bench 4.0: 65.7%
- Hallucination rate down to 8.3%
- Hundreds of parallel subagents
- Open-weight flagship
Providers
Compare API prices across providers. Select a provider to view all charges and conditions.
| $4 | $20 | $0.2 | |
| $2 | $10 | $0.2 |
What $10 buys
- Input
- 2.5M tokens
- Output
- 500K tokens
Estimated from current input and output prices, before discounts.
Benchmarks
- 65.7%
Terminal-Bench 4.0
Claude Opus 5.5: 66.4%; GPT-6 Astra: 58.2%
- 8.3%
Kimi Hallu Bench Hard (statement hallucination rate)
Previous generation: 23.4%; lower is better
- 70.4 points
Financial research (internal test)
Previous generation: 58.8 points
- 57.6 points
Financial modeling (internal test)
Previous generation: 50.4 points
Scores as published by the model provider.
Reasoning
- Default effort
- max
- Enabled by default
- Yes
- Always enabled
- No
- Available effort levels
- max / high / low / none
Quick Start
Drop-in code to call this model with Felo API Platform compatible APIs. Start in the Playground when you want to test prompts before writing code.
Get your API key
Create a key from your Felo API Platform account and set it as an environment variable.
export FELO_API_KEY="YOUR_API_KEY"Make your first request
Use this model ID with the Felo API Platform endpoint. Optional request headers such as X-Request-Id can help troubleshooting.
kimi-k3.1-proThis model is reasoning-capable. If your selected protocol and model route expose reasoning controls, preserve the returned reasoning metadata when continuing a conversation.
const response = await fetch("https://openapi.felo.ai/api/v1/messages", {
method: 'POST',
headers: {
Authorization: `Bearer ${process.env.FELO_API_KEY}`,
'Content-Type': 'application/json',
},
body: JSON.stringify({
"model": "kimi-k3.1-pro",
"max_tokens": 1024,
"messages": [
{
"role": "user",
"content": "Hello"
}
]
}),
});
if (!response.ok) throw new Error(await response.text());
console.log(await response.json());Using third-party SDKs
Use compatible SDKs by setting baseURL to https://openapi.felo.ai/api/v1. Use the Anthropic-compatible base URL for Claude-style clients. https://openapi.felo.ai/api.
Enable streaming
Add stream: true to receive responses as server-sent events.
curl -N https://openapi.felo.ai/api/v1/messages \
-H "Authorization: Bearer $FELO_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "kimi-k3.1-pro",
"max_tokens": 1024,
"messages": [
{
"role": "user",
"content": "Hello"
}
],
"stream": true
}'Endpoint
Felo API Platform exposes compatible model-calling surfaces. Pick the surface your app or agent client already supports.
Messages
Anthropic-compatible messages for Claude-style clients.
POST
https://openapi.felo.ai/api/v1/messagesParameters
modelstringkimi-k3.1-proPublic model ID to call.max_tokensinteger—Maximum generated tokens.messagesarray—Conversation messages for Chat Completions.frequency_penalty———include_reasoning———presence_penalty———reasoning———reasoning_effort———response_formatobject—Request structured output when supported.stop———structured_outputs———tool_choicestring or object—Controls whether a tool should be called.toolsarray—Tool calling definitions following compatible schema.Frequently asked questions
What is MoonshotAI: Kimi K3.1 Pro?
Kimi K3.1 Pro is Moonshot AI's flagship open-weight multimodal reasoning model and the first release in the K3.1 series, launched on October 23, 2026. It is a post-training upgrade of Kimi K3, described by the company as a 3T-class model with a total parameter count reported to match K3's 2.8 trillion. It is offered through the Kimi website, Kimi Work, Kimi Code, and the API. The model has native visual understanding: it accepts image and video input and offers a 1M-token context window for very long text and complex tasks. Compared with the previous generation, coding, 3D creation, and financial research improved notably. On Terminal-Bench 4.0 it scores 65.7%, close to Claude Opus 5.5 at 66.4% and above GPT-6 Astra at 58.2%; internal tests put financial research at 70.4 points, up from 58.8, and financial modeling at 57.6, up from 50.4. Individual-statement hallucination rate on Kimi Hallu Bench Hard dropped from 23.4% to 8.3%. It fits long-horizon coding, 3D model and game design, financial statement analysis and investment research, financial modeling, and end-to-end agentic knowledge work. Multi-agent orchestration scales to hundreds of parallel subagents for autonomous task decomposition. The model is open-weight, with full weights planned for release before November 4, 2026, and its cyber-offense capability is deliberately capped as a safety measure.
How is usage billed?
Your bill depends on the provider you use, its prices and your usage. Longer inputs or other conditions may have different prices.
What inputs does Kimi K3.1 Pro support?
It has native visual understanding and accepts image and video input alongside long text. The model is built to handle very long documents and complex tasks.
How large is the context window?
Kimi K3.1 Pro offers a context window of up to 1M tokens, suited to long documents and multi-step work.
What is Kimi K3.1 Pro best at?
It is aimed at long-horizon coding, 3D model and game design, financial statement analysis and investment research, financial modeling, and end-to-end agentic knowledge work.
How does it differ from Kimi K3?
Kimi K3.1 Pro is a post-training upgrade of Kimi K3 with clearly stronger coding, 3D creation, and financial research results. Its hallucination rate on Kimi Hallu Bench Hard falls from 23.4% to 8.3%.
What is the Highspeed version?
Kimi K3.1 Pro Highspeed keeps the same model capability while running roughly 2 to 3 times faster.
Does it support multi-agent orchestration?
Yes. It supports multi-agent orchestration that can scale to hundreds of parallel subagents for autonomous task decomposition.
