# Kimi K3.1 Pro

> Kimi K3.1 Pro is the first model in Moonshot AI's K3.1 series, a post-trained version of Kimi K3 with a 1M-token context window and native visual understanding. It improves on K3 in coding (Unreal, Unity and Godot projects), 3D and CAD modeling, and financial research, with a lower hallucination rate. Thinking is always on, with low, high and max reasoning effort.

**ID**: `moonshotai/kimi-k3.1-pro`  
**Creator**: Moonshot AI  
**Category**: multimodal  
**Context**: 1M tokens  
**Pricing**: $2.00 / $10.00 per 1M · billed by your provider (BYOK)  
**Released**: 2026-10-23  
**Web page**: https://anyrouter.dev/model/moonshotai/kimi-k3.1-pro

**Input modalities**: text, image, video  
**Output modalities**: text  

**Capabilities**: chat, reasoning, streaming, function-calling, vision, coding, agentic

**Aliases**: `moonshotai/Kimi-K3.1-Pro`

## Usage

**Endpoint**: `POST https://anyrouter.dev/api/v1/chat/completions`

```bash
curl https://anyrouter.dev/api/v1/chat/completions \
  -H "Authorization: Bearer $ANYROUTER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "model": "moonshotai/kimi-k3.1-pro",
  "messages": [
    {
      "role": "user",
      "content": "Say hi in 3 words."
    }
  ]
}'
```

## Providers

| Provider | Input | Output |
| --- | --- | --- |
| moonshotai-byok | $2.00/M (list) | $10.00/M (list) |
| vercel-byok | $2.00/M (list) | $10.00/M (list) |
| requesty-byok | $2.00/M (list) | $10.00/M (list) |
| aihubmix-byok | $2.00/M (list) | $10.00/M (list) |
| openrouter-byok | $2.00/M (list) | $10.00/M (list) |

BYOK routes are billed by your own provider key — AnyRouter charges $0. Rates marked (list) are the provider's published list price.

## Model card

- Architecture: mixture of experts [2]
- Total parameters: 2800B [2]
- Active parameters: 104B [2]
- Native context: 1048576 tokens [2,3]

### Strengths [1]

- Full-project coding across Unity, Unreal Engine and Godot
- Parametric 3D and CAD modeling in Blender and CAD tools
- Lower hallucination on financial research (claim-level rate 8.3% vs 23.4% for K3 on the Hard bench)

### Known limits [3]

- Thinking is always on and cannot be turned off; lower reasoning_effort to cut latency
- Vision input takes base64 or ms:// file ids, not public image URLs
- temperature, top_p, n and the penalty parameters are fixed; omit them

### Use cases [1]

- Long-horizon coding agents
- Financial modeling and research
- 3D and CAD generation from a brief or reference image

### API notes [3]

- reasoning_effort accepts low, high and max (default max)
- max_completion_tokens defaults to 131072 and can be set up to 1048576
- Automatic context caching with 5 minute and 1 hour TTL tiers

### Links

- announcement: https://kimi.ai/blog/kimi-k3-1-pro
- docs: https://platform.kimi.ai/docs/guide/kimi-k3-1-pro-quickstart
- pricing: https://platform.kimi.ai/docs/pricing/chat
- Previous version: https://anyrouter.dev/model/moonshotai/kimi-k3

### Sources

1. [Kimi K3.1 Pro tech blog](https://kimi.ai/blog/kimi-k3-1-pro) (accessed 2026-10-24)
2. [Kimi K3.1 Pro announcement](https://x.com/kimi_moonshot/status/2113680701770068067) (accessed 2026-10-24)
3. [Kimi K3.1 Pro API guide](https://platform.kimi.ai/docs/guide/kimi-k3-1-pro-quickstart) (accessed 2026-10-24)

## Source

- Upstream docs: https://platform.kimi.ai/docs/guide/kimi-k3-1-pro-quickstart