Kimi K3.1 Pro in Europe: run Moonshot's flagship on Verda GPUs in Germany
Bring your own Verda key and Kimi K3.1 Pro runs on GPUs in Germany instead of Moonshot's servers in China, at a 0% fee. Same model id, same code.
Kimi K3.1 Pro is one of the strongest open-weight models you can call today: a 3T-class model with a 1M-token context window, built for coding, 3D and financial agents. Until now there was one catch for a lot of European teams. Calling it through Moonshot's API means your prompts are processed on servers in China.
Today that changes. Moonshot has named Verda as an inference partner for Kimi K3.1 Pro, and Verda runs the model on its own GPUs in Germany. On CompanyFabric you can now route Kimi through Verda with your own Verda key, at a 0% fee, without changing a line of code.
What changes and what doesn't
| Moonshot API | Verda (with your key) | |
|---|---|---|
| Where the model runs | China | Germany |
| Company | Moonshot AI | Verda Cloud Oy, Finland |
| Model | Kimi K3.1 Pro | Kimi K3.1 Pro (same weights) |
| Price per 1M tokens (in / cache read / out) | $2 / $0.20 / $10 | $2 / $0.20 / $10, billed by Verda |
| CompanyFabric fee | Standard | 0% (bring your own key) |
| Your model id | moonshotai/kimi-k3.1-pro | moonshotai/kimi-k3.1-pro |
The model id stays exactly the same. CompanyFabric decides where the request goes based on the keys in your vault, so your application, your agents and your prompts don't need to know Verda exists.
Set it up in two minutes
- Create an API key in the Verda console. Verda bills you directly for the tokens you use.
- Open the BYOK vault [blocked] in your CompanyFabric dashboard and add it under Verda (Kimi K3.1 Pro, EU).
- Call Kimi as you already do:
With Verda as the only Kimi key in your vault, every Kimi K3.1 Pro request runs in Germany. Requests on your own key carry no CompanyFabric fee, so the balance in your dashboard doesn't move.
[Image blocked: Kimi K3.1 Pro on CompanyFabric]
Why bring-your-own-key only
We'd love to offer Verda on CompanyFabric credits too. Verda's terms don't allow reselling its service, so for now the Verda route only works with your own key. We enforce this in the gateway itself: our own traffic, the companyfabric/auto router and our health checks never reach Verda, whatever happens with configuration. If Verda agrees to a reseller arrangement, we'll switch it on and say so in the changelog [blocked].
Your Verda key is encrypted at rest in the vault and is only ever sent to inference.verda.com.
Who this is for
- EU companies with data-residency rules that liked Kimi's quality but couldn't send customer data to China.
- Agencies and consultancies whose client contracts require processing inside the EU.
- Anyone comparing providers. Run the same eval set against Moonshot and Verda by swapping the key in your vault, with no code change.
What's next
Verda also serves DeepSeek V4.1 Flash, and GLM 5.3 is coming soon. Verda's DeepSeek is a later revision than the deepseek/deepseek-v4-1-flash we already serve, so we haven't mapped it onto our id: serving a different model under the same name would make your evals lie. We'll add it as its own model if there's demand.
Read more on the Kimi K3.1 Pro API page [blocked], or see every model we route in the model library [blocked].
- Kimi
- Moonshot AI
- Verda
- BYOK
- Data residency
- EU
Every model in this post is behind one key and one balance. Free starter credits on signup.
Get API KeyKeep reading
New on CompanyFabric: Claude Fable 5.2, Gemini 4 Argon, Muse Wonder, MiMo V2.6, Kimi K3.1 and decision models
Five frontier model families from Anthropic, Google, Meta, Xiaomi and Moonshot, plus decision models from TypeSafe, Perplexity and Microsoft. One key for all of them.
Qwen4-Max is live: Alibaba's new flagship, 1M context, $1.50 per million tokens
Qwen4-Max reads text, images and video, reasons and calls tools across a 1M-token window. Live on CompanyFabric the day it launched, for $1.50 in and $4.50 out.
Claude Code with any model: GPT, Gemini, DeepSeek and Kimi on one key
Set two environment variables and Claude Code runs on any model in the CompanyFabric library, with prompt caching intact on Claude and a budget cap on every key.