ModelSheep

Pricing

Prices per 1M tokens, in USD. You pay for tokens. No subscription, no seat fee, no minimum.

$10 of credit when you sign up

It lands when you verify your email. No card, and the credit is spent at the prices below.

Create account

Compare plans

Every plan runs on the same servers in Germany, at the same prices per token. What changes is how you pay and what we sign.

Free trial

$10 credit

Granted when you verify your email. No card.

Create account

Usage

Credit at signup
$10
Models
Qwen3.5 35B A3B and Qwen3.8 27B
Tokens per minute
500,000
Zero data retention
Servers in Germany

Billing

Data processing agreement
Standard
NDA
Our standard NDA

Support

Support level
Online community and docs

Pay as you go

Per token

Top up credits and spend them on every model we serve.

Create account

Usage

Credit at signup
$10
Models
All models
Tokens per minute
1,000,000
Zero data retention
Servers in Germany

Billing

Pay as you go
Payment methods
Credit card
Payment terms
Fill up credits, Net 0
Billing cycle
Refill as you go
Invoice
A PDF invoice with your billing details is written on every payment, and waits in your account
Data processing agreement
Standard
NDA
Our standard NDA
How to pay
Log in, press Buy credits, enter an amount, give your payment details and press Add credits

Support

Support level
Level 2
Availability
Mon to Fri, 9:30 to 16:00 ET
Response time
Best effort
Email support

Enterprise

Custom

Volume, a contract, and a model of your own.

Contact us

Usage

Credit at signup
Custom
Models
All models, and your own
Tokens per minute
40,000,000
Zero data retention
Servers in Germany
Bring your own model

Billing

Pay as you go
Yes, and a contract
Payment methods
Credit card, wire transfer
Payment terms
Custom
Billing cycle
Custom
Invoice
Custom
Purchase orders
Due diligence questionnaires
Onboarding in your ERP
Terms
Custom
Data processing agreement
Custom
NDA
Custom, on request
How to pay
Custom

Support

Support level
Level 1
Availability
Mon to Fri, 9:30 to 16:00 ET
Response time
Within one business day
Email support
Phone support
Onboarding and premium support

Price per model

Language Models

Text to text.

ModelContextInputOutputCheapest elsewhere
GLM 5.3zai-org/GLM-5.31.3M$0.915$2.875$1.017 / $3.195
Kimi K3moonshotai/Kimi-K31M$2.250$11.475$2.500 / $12.750
GLM 5.3 Flashzai-org/GLM-5.3-Flash1.3M$0.081$0.270$0.090 / $0.300
Qwen3.8 2.4T A95BQwen/Qwen3.8-2.4T-A95B1M$1.800$5.400$2.000 / $6.000
DeepSeek V4.1 Flashdeepseek-ai/DeepSeek-V4.1-Flash1M$0.135$0.540$0.150 / $0.600
DeepSeek V4 Pro 0813deepseek-ai/DeepSeek-V4-Pro-08131M$0.522$1.568$0.581 / $1.742
DeepSeek V4 Flash 0731deepseek-ai/DeepSeek-V4-Flash-07311.3M$0.027$0.117$0.030 / $0.130
Qwen3.8 27BQwen/Qwen3.8-27B-FP81M$0.180$1.980$0.200 / $2.200
DeepSeek V4 Prodeepseek-ai/DeepSeek-V4-Pro1M$0.939$1.879$1.044 / $2.088
Minimax M3MiniMaxAI/Minimax-M31M$0.252$0.990$0.280 / $1.100
Kimi K2.6moonshotai/Kimi-K2.6262K$0.513$2.160$0.570 / $2.400
Qwen3.5 35B A3BQwen/Qwen3.5-35B-A3B262K$0.126$0.900$0.140 / $1.000

Vision Models

OCR and image to text.

ModelContextInputOutputCheapest elsewhere
GLM OCRzai-org/GLM-OCR8K$0.027$0.027$0.030 / $0.030

Embedding Models

Text to vector, for search and retrieval.

ModelContextInputOutputCheapest elsewhere
Qwen3 Embedding 8BQwen/Qwen3-Embedding-8B33K$0.009n/a$0.010 / n/a
Qwen3 Embedding 4BQwen/Qwen3-Embedding-4B33K$0.009n/a$0.010 / n/a
Qwen3 Embedding 0.6BQwen/Qwen3-Embedding-0.6B33K$0.004n/a$0.005 / n/a
BGE M3BAAI/bge-m38K$0.009n/a$0.010 / n/a
Nemotron 3 Embed 8Bnvidia/Nemotron-3-Embed-8B-BF1633K$0.031n/a$0.035 / n/a
EmbeddingGemma 300Mgoogle/embeddinggemma-300m2K$0.001n/a$0.002 / n/a

Audio Models

Transcribe, speech to text and text to speech.

ModelContextInputOutputCheapest elsewhere
Voxtral Small 24Bmistralai/Voxtral-Small-24B-250733K$0.090$0.270$0.100 / $0.300
MiMo V2.5XiaomiMiMo/MiMo-V2.51.1M$0.126$0.252$0.140 / $0.280
Faster Whisper Large v3Systran/faster-whisper-large-v30K$0.024 /hrn/a$0.027 / n/a

Every price is 10 percent under the lowest standing price for the same model elsewhere, taken per input token and per output token. Promotional prices are counted at their list price, and providers serving a lower precision than the FP8 we serve are not counted. Checked on 2026-09-17.

Questions

How do I pay?
The $10 credit lands when you verify your email, so the first calls need no payment at all. After that: log in, press Buy credits, enter an amount, give your payment details and press Add credits.
Which models does the free trial reach?
The trial credit is spent on Qwen3.5 35B A3B and Qwen3.8 27B. Every other model opens as soon as you top up.
What happens when the credit runs out?
Every call answers 402 until the balance is above zero again. Nothing is queued and nothing is lost. Your keys keep working, so a top up puts you back to where you were.
Do the prices change?
The promise is 10 percent under the lowest standing price for the same model elsewhere, checked per input token and per output token. The prices above were checked on 2026-09-17.
Is there a minimum or a monthly fee?
No. There is no subscription, no seat fee and no minimum. The price is the same for one request a day and for a million.
Do you keep my prompts?
No. A prompt and an answer are held in memory for the length of the request and are not written to disk. We count tokens, because the call is priced on them, and nothing else.