Model comparison · 2026-10-10
Can You Use Kimi K3 Through an EU-Hosted API?
Call Kimi K3 on LLMTR as tensorx/kimi-k3-tensorx: processed in the EU per the provider, a data-transfer consent is required, and prices are listed.
Short answer: yes, with tensorx/kimi-k3-tensorx
LLMTR (llmtr.com) is a gateway that gives access to language models hosted in Türkiye and at global providers through one OpenAI-compatible API. You reach Kimi K3 at https://llmtr.com/v1 with the model id tensorx/kimi-k3-tensorx. The row is served by TensorX and works with any OpenAI-compatible SDK.
For the models in this category, processing inside the European Union is what the provider states; LLMTR has not independently verified it. According to the provider's data processing agreement, this row is processed inside the EU and inputs are not used for model training.
To call it, the TensorX data-transfer consent must be granted on your account. Without it every request returns 403 provider_consent_required; the request does not reach TensorX and is not billed.
Can I use Kimi K3 through an EU-hosted API?
The tensorx/kimi-k3-tensorx row has a 1,048,576-token context, accepts image input, and supports tool calling and streaming (SSE). The price per 1M tokens is $3.00 input, $0.75 for input read from cache, and $15.00 output. These are TensorX's own USD list prices, and LLMTR adds no markup to the token price.
The cache price applies only when the provider reports tokens read from its cache; otherwise input is billed at the normal input rate. No cache hit rate is promised.
Where data goes: the Moonshot platform and the TensorX row
The two access paths do not follow the same data path. The privacy policy of Moonshot AI's own platform (last update 30 Apr 2025) says: "We store the information we collect in secure servers located in Singapore." That statement is Moonshot's, written for its own platform; it does not describe a call to the TensorX row through LLMTR.
For the TensorX row, the information is the provider's statement: requests are processed inside the European Union. For the models in this category, processing inside the European Union is what the provider states; LLMTR has not independently verified it.
| Access path | Where data is processed (per the stated source) | Source |
|---|---|---|
| Moonshot AI's own platform | Collected information is stored on servers in Singapore | Moonshot AI privacy policy (last update 30 Apr 2025) |
| tensorx/kimi-k3-tensorx through LLMTR | Processed inside the European Union, per the provider's statement | TensorX data processing agreement (DPA) |
Cost and what to watch on the first call
Worked example: 1M input tokens and 100K output tokens, without cache, cost $3.00 + $1.50 = $4.50. Kimi K3 reasons by default, and reasoning tokens are billed as output. If a reply ends with finish_reason: length, raise max_tokens and retry.
reasoning_effort is not supported on this row; a request that carries it is refused with 400 invalid_request, not sent and not billed. Image input is supported. For usage in streams, send include_usage as true inside stream_options. The current price is always on the model page.
On endpoints: chat calls go to /v1/chat/completions, and TensorX chat rows also answer on /v1/messages for Anthropic-format clients. /v1/responses is not supported on these rows and returns 400 unsupported_operation. The same API key and the same credit balance work for every model in the catalog, so switching models means changing the model string. You can also limit a key to selected models and set a spending limit.
Example call and a fallback model
The code below calls tensorx/kimi-k3-tensorx with one OpenAI-compatible client. Kimi K3 has no same-family pair in the catalog. If you need another model, pick one from the EU-hosted category by need; the Kimi K2.7 Code and GLM-5.3 families, for example, are on both GreenPT and TensorX. Test the model you pick on your own prompts and grant its provider's data-transfer consent. LLMTR does not switch providers automatically; catching the error and calling another model id is the client's job.
The GDPR assessment stays with you. LLMTR and the provider's statement give input to that assessment and do not replace its result. The full GDPR text and the TensorX DPA are linked as sources. You decide which data is sent, which legal basis applies, and how records are kept. Classify your own data first, then choose the model id that fits that class and limit the API key to those ids.
Python example calling tensorx/kimi-k3-tensorx with max_tokens 2048.
import os
from openai import OpenAI
client = OpenAI(base_url="https://llmtr.com/v1", api_key=os.environ["LLMTR_API_KEY"])
# Kimi K3 reasons by default: reasoning tokens are billed as output.
# If finish_reason is "length", raise max_tokens and retry.
# Do not send reasoning_effort: it is refused with 400 invalid_request.
reply = client.chat.completions.create(
model="tensorx/kimi-k3-tensorx",
messages=[{"role": "user", "content": "Summarize this paragraph: ..."}],
max_tokens=2048,
)
print(reply.choices[0].finish_reason)
print(reply.choices[0].message.content)Getting started on LLMTR: account, consent, key and first request
Step 1: create an account at llmtr.com, verify your email, and top up credit from the Dashboard. An 8% platform margin is added once on the top-up; a $10.00 credit is charged $10.80. Each call is then deducted at the model's catalog price.
Step 2: in Dashboard > Settings, open "Data-transfer consents" and grant the TensorX consent. It is account-level, covers every API key, and can be withdrawn in the same place. What is transferred is the request content, to produce the answer; name and email are not. A GreenPT consent does not cover TensorX.
Step 3: create a key under Dashboard > API Keys; the raw key is shown once, so store it as LLMTR_API_KEY. Step 4: send the first request to https://llmtr.com/v1 with the example above. LLMTR does not store prompt and response bodies.
Call Kimi K3 through LLMTR
Create an account, top up credit, grant the TensorX data-transfer consent, create an API key, and send the first request to tensorx/kimi-k3-tensorx.
- Account and credit. Create an account at llmtr.com, verify your email, and top up credit from the Dashboard. An 8% margin is added once; a $10.00 credit is charged $10.80.
- TensorX data-transfer consent. In Dashboard > Settings, grant the TensorX consent under "Data-transfer consents". Without it, requests return 403 provider_consent_required.
- API key. Create a key under Dashboard > API Keys. The raw key is shown once; store it as the LLMTR_API_KEY environment variable.
- First request. Send the first request for tensorx/kimi-k3-tensorx to https://llmtr.com/v1 with any OpenAI-compatible SDK, changing only base_url and api_key.
Frequently asked questions
Can I use Kimi K3 through an API that processes data in the EU?
LLMTR has the tensorx/kimi-k3-tensorx row. Per the provider's statement, requests are processed inside the European Union; LLMTR has not independently verified it. The TensorX data-transfer consent is required to call it.
Is Kimi K3 GDPR compliant?
That is an assessment of your use, not of a model, and it stays with you. The provider's statement and the TensorX DPA are input to your assessment and do not replace its result.
What happens if I do not grant the consent?
The request returns 403 provider_consent_required, does not reach TensorX, and is not billed. You grant the consent in Settings under "Data-transfer consents".