Kimi K3 API
Kimi K3 is a text model by Moonshot. On Nodion you call it through the OpenAI-compatible AI Gateway with the model id moonshot/kimi-k3.
- Input
- from 3,00 € per 1M tokens
- Output
- 15,00 € per 1M tokens
- Cached input
- from 0,30 € per 1M tokens
Run Kimi K3 through our EU AI Gateway
One endpoint in front of every provider, operated in the EU.
- Failover across 4 providers If a provider cannot be reached, the next requests go to the others your project allows.
- No provider accounts Your Nodion key calls every provider, and you pay for what you use from a prepaid wallet or through an AI Coding Plan.
- EU and zero data retention Every provider is based in the EU and serves the models with zero data retention. You can select and deselect providers yourself.
Prices in EUR excl. VAT
Providers
Our gateway serves Kimi K3 through the following providers, all of them EU providers with zero data retention. You need no account with the providers: you call everything centrally through Nodion with your Nodion API key.
| Input3,00 € | Output15,00 € | Cached input0,75 € | Latency | Throughput | Uptime | |
| Input3,00 € | Output15,00 € | Cached input0,90 € | Latency | Throughput | Uptime | |
| Input3,00 € | Output15,00 € | Cached input0,75 € | Latency | Throughput | Uptime | |
| Input5,00 € | Output15,00 € | Cached input0,30 € | Latency | Throughput | Uptime |
Model details
- Model id
moonshot/kimi-k3- Input
- from 3,00 € per 1M tokens
- Output
- 15,00 € per 1M tokens
- Cached input
- from 0,30 € per 1M tokens
- Context
- 1M tokens
- Max output
- 32K tokens
- Input / output
- Features
- Reasoning effort
- low, high, max
- Privacy
- ZDR, EU only
- Coding plans
- Available
- Model card
- Hugging Face
Call it
One OpenAI-compatible endpoint.
POST /v1/chat/completions on https://api.nodion.ai. Requests, limits and error codes are in the AI API reference.
curl https://api.nodion.ai/v1/chat/completions \
-H "Authorization: Bearer $NODION_AI_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "moonshot/kimi-k3",
"messages": [{"role": "user", "content": "Hello"}]
}' Every provider on the gateway is an EU provider with zero data retention. In your project's AI settings you can also restrict which providers and regions may serve requests. Details are in the AI Inference docs.
Questions and answers.
Where does Kimi K3 run?
In the EU. Our gateway sends requests to Lyceum, Berget AI, TensorX and Arkane, all of them EU providers; the catalog lists datacenters in Sweden.
Do the providers store my Kimi K3 prompts?
No. Every provider on the Nodion AI Gateway runs with zero data retention and does not store your prompts or outputs.
How much does Kimi K3 cost?
Billing is per 1M input and output tokens and cached input. The current prices per provider are at the top of this page; you pay from a prepaid wallet or through an AI Coding Plan.
How do I call Kimi K3?
With any OpenAI-compatible SDK or tool on POST /v1/chat/completions: base URL https://api.nodion.ai/v1, model id moonshot/kimi-k3 and an API key from the dashboard.
How large is the Kimi K3 context window?
1M tokens, with up to 32K tokens per reply.
Is Kimi K3 included in the AI Coding Plans?
Yes. With an AI Coding Plan you call Kimi K3 on the same endpoint, for example from OpenCode. Usage counts against the plan's windows; when one is full, requests continue only if the plan allows wallet use.
Give your agent a cloud.
Earn up to 15 € in credit when signing up.
or with email
Stay in the loop.
Product news, no spam. Unsubscribe anytime.
Successfully subscribed.