AI · AI Tokens
173 AI models, one API.
Call open and proprietary models through an OpenAI-compatible endpoint and pay only for the tokens you use — from $0.0019 per million input tokens. No GPUs to rent, no servers to run.
Pay-per-token models.
A pick from each provider — the full catalogue is below.
Drop-in OpenAI compatibility
Point an OpenAI SDK at your endpoint’s base URL and keep your code.
Open and proprietary models
Chat, reasoning, coding, vision and embedding models in one catalogue.
Pay only for tokens
Separate input and output rates, deducted from your balance as you go.
Keys you control
Each endpoint has its own API key, which you can rotate at any time.
The model catalogue
Rates per million tokens at the cheapest provider serving each model, deducted as you go.
| Model | Input / 1M | Output / 1M | Served from | Create endpoint |
|---|---|---|---|---|
| all MiniLM L12 v2 Embedding | $0.00475 | — | | Create endpoint |
| all MiniLM L6 v2 Embedding | $0.00475 | — | | Create endpoint |
| all mpnet base v2 Embedding | $0.00475 | — | | Create endpoint |
| AutoGLM Phone 9B Multilingual | $0.03325 | $0.1311 | | Create endpoint |
| Baichuan M2 32B | $0.0665 | $0.0665 | | Create endpoint |
| bge base en v1.5 Embedding | $0.00475 | — | | Create endpoint |
| bge en icl Embedding | $0.0095 | — | | Create endpoint |
| bge large en v1.5 Embedding | $0.0095 | — | | Create endpoint |
| bge m3 Embedding | $0.0095 | — | | Create endpoint |
| bge m3 multi Embedding | $0.0095 | — | | Create endpoint |
| bge m3 multi 8k Embedding | $0.0095 | — | | Create endpoint |
| claude fable 5 | $9.50 | $47.50 | | Create endpoint |
Excl. VAT · Taxes may apply · The same model may be offered by more than one provider.
How it works
From picking a model to your first response in three steps.
Pick a model
Choose from the catalogue by capability and price.
Create an endpoint
The console gives you its base URL and API key.
Call it from your code
Use any OpenAI-compatible SDK with the model’s name.
from openai import OpenAI
client = OpenAI(
base_url="<your endpoint's base URL>",
api_key="<your endpoint's API key>",
)
reply = client.chat.completions.create(
model="<model>",
messages=[{"role": "user", "content": "Hello!"}],
)
print(reply.choices[0].message.content)What you can build
What teams build on pay-per-token models.
Chat and assistants
Power support bots and internal assistants with a hosted LLM.
Search and RAG
Embed documents and retrieve the right context for each answer.
Coding tools
Wire code-tuned models into editors, CI checks and review bots.
Agents and automation
Let agents call models for planning, extraction and tool use.
Translation and summaries
Process documents, tickets and transcripts at a per-token price.
Prototyping
Try several models side by side before committing to one.
Supported providers
2 providers in 1 country offer AI models today, with more joining as the network grows.
Need your own GPUs?
When a hosted API is not enough, run models on dedicated GPUs.
Frequently asked questions
What teams ask about AI Tokens.
Make your first call in minutes.
Pick a model, create an endpoint, and pay only for the tokens you use.