Models
Our models
All models you can call with a TOKENWERK API key – compatible with the OpenAI and Anthropic APIs. Billing is per call, by the tokens used.
Every model has its own TOKENWERK name, often after its purpose. You choose what you need it for. We choose, test and maintain what answers behind it, and can improve it without you changing your code. The “Kind” column shows how a call finds its target: Direct uses fixed routes by priority; if a provider fails, the next one takes over. Auto picks the right model per call (in progress).
Loading models …
No models are available yet.
The models could not be loaded right now. Please try again in a minute.
| Model | Kind | Context window | Max output | Input | Output | Cached input |
|---|
Prices in euros per 1 million tokens, billed per call from your organization's balance.
Use a model
Call it with an API key of your organization (it starts with tw-), e.g. from the environment
variable TOKENWERK_API_KEY.
Examples for more SDKs and for agents such as opencode, OpenClaw or Hermes are in the docs.
Auto models choose the model for each call. Automatic routing is not available yet, calls to this model are rejected for now.
Save it as
opencode.json in your project (or globally as ~/.config/opencode/opencode.json)
and start opencode. Install: npm install -g opencode-ai.