Open source model API

Zero data retention.
Up and running in less than 60 seconds.

from openai import OpenAI
 
client = OpenAI(
    base_url="https://api.modelsheep.com/v1",
    api_key=API_KEY_MODELSHEEP,
)
 
response = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-V3",
    messages=[{
      "role": "user", "content": "Why do sheep count people?"
    }],
)

Swap models without swapping hosts.

Ordered by usage on OpenRouter. Copy a model ID into your request.

Model	Contexttokens	InputUSD per 1M tokens	OutputUSD per 1M tokens
DeepSeek V3	128K	$0.14	$0.28
Qwen3 235B	256K	$0.12	$0.50
Kimi K2	128K	$0.50	$2.00
Llama 3.3 70B	128K	$0.10	$0.28
GLM 4.6	200K	$0.40	$1.75
Qwen3 Coder 480B	256K	$0.30	$1.20
DeepSeek R1	128K	$0.50	$2.15
gpt-oss 120B	128K	$0.09	$0.45
Mistral Small 3.2	128K	$0.05	$0.10
Llama 4 Maverick	1M	$0.17	$0.60

See the full model list

You do not pay extra for Germany.

You pay for tokens. No subscription, no seat fee, no minimum. The price is the same for one request a day or a million.

One million tokens in and one million tokens out on DeepSeek V3.

modelsheep

$0.42

Typical US host

2.1x modelsheep$0.90

Typical EU host

3.0x modelsheep$1.25

Public list prices, checked on 2026-09-16.

When the response ends, your prompt is gone.

Zero data retention

We hold prompts and completions in memory only. We never write them to disk. We keep the token count for billing. We keep nothing else.

Made in Germany

The API runs on servers in Germany. We process your requests there and nowhere else. No third country gets access to them.

DSGVO

We act as your processor under the DSGVO. You stay the controller. You can sign the data processing agreement before your first request.

Read the data processing agreement

You change two lines.

1
Create your API key.
One form. No sales call.
2
Set the base URL.
https://api.modelsheep.com/v1
3
Send the first request.
Your prompts stay in Germany.

Get API key