---
title: "Kimi K3 API hosted in the EU"
description: "Run Kimi K3, Moonshot AI's 2.8-trillion-parameter open-weight model with 1M context and image and video input, on inference.eu with processing in the EU."
canonical: "https://inference.eu/models/kimi-k3/"
language: "en"
---

Reasoning · Moonshot AI

# Run Kimi K3 in the EU through an OpenAI-compatible API

Kimi K3 is Moonshot AI's most capable model and the first open-weight model with 2.8 trillion parameters. It understands text, images and video and handles a one-million-token context. On inference.eu you call it with the OpenAI or Anthropic request format, processed in EU data centres and never used for training.

[Contact sales](https://inference.eu/contact/)[Example request](#example)

1.  012.8T · 104B
2.  02EU data centres
3.  03Kimi K3 License

## What Kimi K3 is

Kimi K3 launched in July 2026, with full weights published on 27 July. It is a mixture-of-experts model with 2.8 trillion parameters, 104 billion of them active per token: each token is routed to 16 of 896 experts. It has 93 layers, most of which use Moonshot's Kimi Delta Attention, mixed with gated latent attention layers.

Moonshot says the new architecture gives about 2.5 times the scaling efficiency of Kimi K2. The model reads text, images and video natively and has a one-million-token context window. It always reasons, with effort levels low, high and max, and was trained with quantisation-aware training in MXFP4 weights. It is released under its own Kimi K3 licence.

## What it does well, and where it falls short

Kimi K3 is aimed at the hardest open-model workloads: long engineering sessions on large repositories, deep research with many tool calls, and knowledge work that mixes documents, images and video. Its one-million-token window lets an agent keep huge amounts of material in view. For tasks where a mid-sized model keeps failing, K3 is a strong escalation step.

It is an extremely large model, so it costs more per token than almost anything else in our catalogue, and its default max reasoning produces long outputs. For everyday tasks, Kimi K2.6, GLM-5.3-Flash or DeepSeek V4.1 Flash are more economical. Its training reflects its maker's choices on sensitive topics, and its licence has conditions worth reading.

## Calling Kimi K3 through the API

inference.eu offers the OpenAI and the Anthropic request formats. Set the base URL to the inference.eu API, use your inference.eu key and set the model to `kimi-k3`. Coding agents and research tools that accept a custom endpoint work without further changes.

Kimi K3 always thinks and returns `reasoning_content` alongside the answer. Set `reasoning_effort` to `low`, `high` or `max`; the default is `max`. K3 was trained with preserved thinking history, so in multi-turn conversations and tool loops pass each full assistant message back unchanged, including its reasoning. Send images as `image_url` parts. Moonshot uses `temperature` 1.0 in its evaluations.

## Data residency for Kimi K3

Kimi K3 is developed in China, but on inference.eu the open weights run on infrastructure in the European Union. Your prompts, images, video and the model's responses are processed in EU data centres, are not sent to Moonshot AI, and are never used to train any model.

Every customer gets a data processing agreement under Article 28 of the GDPR and our subprocessor list. Billing is in euros. The Kimi K3 licence permits use, modification and sale. Products with more than 100 million monthly active users or 20 million US dollars of monthly revenue must display "Kimi K3" prominently, and internal use that exposes nothing to third parties is exempt from the extra conditions.

## Kimi K3

Maker

Moonshot AI

Made in

China

Type

Reasoning

Parameters

2.8T

Active parameters

104B

Context window

1,048,576 tokens

Licence

[Kimi K3 License](https://huggingface.co/moonshotai/Kimi-K3/blob/main/LICENSE)

Released

July 2026

Model ID

kimi-k3

Processing

EU data centres

Training on your data

Never

Source

[huggingface.co](https://huggingface.co/moonshotai/Kimi-K3)

kimi-k3 · Python

```
import os
from openai import OpenAI

client = OpenAI(base_url="https://api.inference.eu/v1", api_key=os.environ["INFERENCE_EU_API_KEY"])

response = client.chat.completions.create(
    model="kimi-k3",
    messages=[{"role": "user", "content": "Summarise this contract in five bullet points."}],
)
print(response.choices[0].message.content)
```

## What teams use it for

-   1

    ### Hard engineering tasks

    Escalate difficult debugging, performance work or large migrations to K3 when smaller models stall.

-   2

    ### Deep research with media

    Combine reports, screenshots and recorded video in one long-context analysis with many tool calls.

-   3

    ### Repository-wide understanding

    Load a very large codebase into the one-million-token window and ask architectural questions about it.

## Kimi K3: common questions

The questions buyers, developers and legal teams ask us most. Something missing? Ask us directly.

Compliance depends on how you process data. On inference.eu, Kimi K3 requests are processed in EU data centres, not used for training and covered by a data processing agreement under Article 28 GDPR with a published subprocessor list. Nothing is sent to Moonshot AI or to China.

On inference.eu the open weights run in EU data centres, so prompts, images and video do not leave the EU for inference. The model's answers still reflect its training, including on politically sensitive topics. Test it on your own tasks before relying on it in production.

Kimi K3 has 2.8 trillion parameters in total, with 104 billion active per token. Each token is routed to 16 of 896 experts. Moonshot describes it as the world's first open 3T-class model and the largest open-weight model released to date.

Kimi K3 supports a one-million-token context window. That is enough for very large codebases, long research corpora or hours of transcribed material. Leave room in the output budget for reasoning, which is always on.

Yes, under the Kimi K3 licence. You may use, modify and sell. Products above 100 million monthly active users or 20 million US dollars of monthly revenue must display Kimi K3 prominently. Large model-as-a-service operators need a separate agreement with Moonshot; that clause applies to the provider, not to you as an API user.

Yes. Kimi K3 is natively multimodal and understands text, images and video in the same model. Send images in the standard \`image\_url\` format. It produces text output, so it can describe and analyse media but does not generate it.

Kimi K3 was trained with preserved thinking history. In multi-turn chats and tool loops, send each complete assistant message back as it was returned, including \`reasoning\_content\`. Dropping the reasoning can make later turns less reliable.

Both are large open models for long-horizon coding with one-million-token context. Kimi K3 is far larger, reads images and video, and is stronger on mixed-media work. GLM-5.3 is text-only and lighter. Their licences differ, so check both against how you plan to use them.

Set your OpenAI client's base URL to the inference.eu API, use your inference.eu key and set the model to \`kimi-k3\`. Keep assistant messages intact in your history, including reasoning, and set \`reasoning\_effort\` to control depth and cost.

No. Moonshot AI published the weights but takes no part in serving your requests on inference.eu. Prompts and media are processed in EU data centres, used only for your answer and never used to train Kimi K3 or any other model.

[Ask a question](https://inference.eu/contact/)

## Related

[All models](https://inference.eu/models/#directory)

-   [Kimi K2.6Run Moonshot AI's Kimi K2.6 on inference.eu: a 1-trillion-parameter open agentic model with image input and 256K context, processed in EU data centres.Reasoning](https://inference.eu/models/kimi-k2-6/)
-   [GLM-5.3Run Z.ai's GLM-5.3, an open-weight model built for complex coding and long agent tasks, on inference.eu with OpenAI and Anthropic APIs, processed in the EU.Reasoning](https://inference.eu/models/glm-5-3/)
-   [DeepSeek V4.1 Flash (1017)Run DeepSeek V4.1 Flash (1017) on inference.eu: a multimodal open reasoning model with 1M context, OpenAI-compatible, processed in the EU, not in China.Reasoning](https://inference.eu/models/deepseek-v4-1-flash-1017/)
-   [MiniMax M3Run MiniMax M3 on inference.eu: an open multimodal model with 1M context for coding and agent work, OpenAI-compatible and processed in EU data centres.Reasoning](https://inference.eu/models/minimax-m3/)

## Move your AI workloads to EU infrastructure

Tell us which models you use and roughly how much traffic you have. We reply within one working day with pricing and next steps.

-   Reply within one working day
-   Pricing for your volume
-   DPA ready to sign