API für Open-Source-Modelle

Das schafste von allen.

Keine Datenspeicherung.
In unter 60 Sekunden einsatzbereit.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.modelsheep.com",
    api_key=API_KEY_MODELSHEEP,
)

response = client.chat.completions.create(
    model="zai-org/GLM-5.3",
    messages=[
        {
            "role": "user",
            "content": "Why do sheep count people?"
        }
    ],
)

print(response.choices[0].message.content)

Offene Modelle

Aktuelle Modelle mit offenen Gewichten und veröffentlichten Kennzahlen. Kein Anbieter auf OpenRouter ist günstiger, weder pro Eingabe-Token noch pro Ausgabe-Token.


GLM 5.3	text	text	753.3B	44.9, 74.8	FP8	1.3M	$0.915	$2.875
Kimi K3	text, image, video	text	2.8T, 104B	43.8, 76.2	FP8	1M	$2.250	$11.475
GLM 5.3 Flash	text, image	text	321.3B, 18B	41.9, 71.5	FP8	1M	$0.081	$0.270
Qwen3.8 2.4T A95B	text	text	2.4T, 95B	40, 71.9	FP8	1M	$1.800	$5.400
DeepSeek V4.1 Flash	text, image	text	552B, 16B	39.5	MXFP4/MXFP8	1M	$0.135	$0.540
DeepSeek V4 Pro 0813	text	text	1.7T, 49B	36.3, 68.8	FP8	1M	$0.522	$1.568
DeepSeek V4 Flash 0731	text	text	284B, 13B	34.5, 69.1	FP8	1M	$0.027	$0.117
Qwen3.8 27B	text, image, video	text	27.8B	33.9, 68.1	FP8	262K	$0.180	$1.980
DeepSeek V4 Pro	text	text	1.6T, 49B	30.9, 59.4	FP8	1M	$0.939	$1.879
Minimax M3	text, image, video	text	427B, 23B	29.6, 58.6	FP8	1M	$0.252	$0.990
Kimi K2.6	text, image	text	1T, 32B	27.5, 61.8	FP8	262K	$0.513	$2.160
Qwen3.5 35B A3B	text, image, video	text	36B, 3B	19.3, 37	FP8	262K	$0.126	$0.900
MiMo V2.5	text, audio, image, video	text	310.8B, 15B	22.3, 56.8	FP8	1.1M	$0.126	$0.252
GLM OCR	text, image	text	0.9B	not rated	FP8	not published	$0.027	$0.027
Qwen3 Embedding 8B	text	vector	7.6B	not rated	FP8	33K	$0.009	n/a
Qwen3 Embedding 4B	text	vector	4B	not rated	FP8	33K	$0.009	n/a
Qwen3 Embedding 0.6B	text	vector	0.6B	not rated	FP8	33K	$0.004	n/a
BGE M3	text	vector	0.568B	not rated	FP8	8K	$0.009	n/a
Nemotron 3 Embed 8B	text	vector	7.953B	not rated	FP8	33K	$0.031	n/a
EmbeddingGemma 300M	text	vector	0.308B	not rated	FP8	2K	$0.001	n/a
Voxtral Small 24B	text, audio, file	text	24.3B	not rated	FP8	33K	$0.090	$0.270
Faster Whisper Large v3	audio	text	1.55B	not rated	FP16	0K	$0.024	n/a

See the full model list

Preise

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 GLM 5.3, the model with the most providers on OpenRouter.

ModelSheep

$3.79

Cheapest of 26 on OpenRouter

$4.21

1.1x ModelSheep

Median provider

$5.80

1.5x ModelSheep

Our price is 10 percent under the cheapest paid provider, on this model and on every other. Checked on 2026-09-17.

At least 10 percent under OpenRouter, on every model.

Each pair shows the cheapest OpenRouter offer and our price for the same model. The rule applies to input tokens and output tokens with no exceptions.

Model	Input per 1M tokens		Output per 1M tokens		You save
Model	OpenRouter	ModelSheep	OpenRouter	ModelSheep	You save
GLM 5.3	$1.017	$0.915	$3.195	$2.875	10%
Kimi K3	$2.500	$2.250	$12.750	$11.475	10%
GLM 5.3 Flash	$0.090	$0.081	$0.300	$0.270	10%
Qwen3.8 2.4T A95B	$2.000	$1.800	$6.000	$5.400	10%
DeepSeek V4.1 Flash	$0.150	$0.135	$0.600	$0.540	10%
DeepSeek V4 Pro 0813	$0.581	$0.522	$1.742	$1.568	10%
DeepSeek V4 Flash 0731	$0.030	$0.027	$0.130	$0.117	10%
Qwen3.8 27B	$0.200	$0.180	$2.200	$1.980	10%
DeepSeek V4 Pro	$1.044	$0.939	$2.088	$1.879	10%
Minimax M3	$0.280	$0.252	$1.100	$0.990	10%
Kimi K2.6	$0.570	$0.513	$2.400	$2.160	10%
Qwen3.5 35B A3B	$0.140	$0.126	$1.000	$0.900	10%
Voxtral Small 24B	$0.100	$0.090	$0.300	$0.270	10%
MiMo V2.5	$0.140	$0.126	$0.280	$0.252	10%
GLM OCR	$0.030	$0.027	$0.030	$0.027	10%
Qwen3 Embedding 8B	$0.010	$0.009	n/a	n/a	10%
Qwen3 Embedding 4B	$0.010	$0.009	n/a	n/a	10%
Qwen3 Embedding 0.6B	$0.005	$0.004	n/a	n/a	20%
BGE M3	$0.010	$0.009	n/a	n/a	10%
Nemotron 3 Embed 8B	$0.035	$0.031	n/a	n/a	11%
EmbeddingGemma 300M	$0.002	$0.001	n/a	n/a	50%
Faster Whisper Large v3	$0.027	$0.024	n/a	n/a	11%

Mit der Antwort ist Ihr Prompt weg.

Keine Datenspeicherung

Prompts und Antworten liegen nur im Arbeitsspeicher. Auf die Festplatte schreiben wir sie nie. Für die Abrechnung bleibt die Zahl der Token. Sonst bleibt nichts.

Made in Germany

Die API läuft auf Servern in Deutschland. Dort verarbeiten wir Ihre Anfragen, und sonst nirgends. Kein Drittland erhält Zugriff darauf.

DSGVO

Wir sind Ihr Auftragsverarbeiter nach DSGVO. Verantwortlich bleiben Sie. Den Auftragsverarbeitungsvertrag können Sie vor der ersten Anfrage unterschreiben.

Sie ändern zwei Zeilen.

1
API-Schlüssel erstellen.
Ein Formular. Kein Vertriebsgespräch.
2
Base URL eintragen.
https://api.modelsheep.com
3
Erste Anfrage senden.
Ihre Prompts bleiben in Deutschland.

API-Schlüssel holen