Skip to main content
The model is the brain doing the thinking inside your framework. You pick one at deploy time and can change it later.
Agents need tool-calling. A model that can’t call tools can’t run the agent loop, so the picker only offers models verified to work end to end.
At launch, inference is routed through Voight’s own OpenRouter key — you don’t bring a key, and these families are subsidized so getting started is cheap (or free with your first agent). Two families are available:

GLM — by Z.ai

The GLM lineup (4.5, 4.5 Air, 4.6 and up). Strong general reasoning and tool use.

MiMo — by Xiaomi

Fast, inexpensive, tool-capable, with a very large context window.
The launch default is MiMo v2.5 — cheap, quick, tool-capable, and comfortable with long context. It’s a sensible starting brain for most agents; switch to a GLM model if you want more reasoning depth.

​
Bring your own key — coming soon

The next phase opens the model choice wide open, with cost billed to your own provider account:
  • BYOK · OpenRouter — paste your OpenRouter key and pick any model in the OpenRouter catalog.
  • BYOK · Direct — connect a provider key directly: OpenAI, Anthropic, Gemini, Qwen, Z.ai, or Xiaomi. Calls go straight to the vendor.
Per-model cost controls land alongside token metering in the same phase.

​
Choosing

You want…Pick
The cheapest capable defaultMiMo v2.5 (launch default)
More reasoning depthA GLM model
A specific frontier modelWait for BYOK, or tell us what you need

​
Next