200+ models worldwide
One way to call them
Seven model types from leading vendors worldwide: text chat, multimodal understanding, structured decision (such as TypeSafe Jev), image generation, video generation, speech, and embeddings & reranking. Every model uses the same API and keys — choose by task, with new models added as they launch.
Leading model vendors worldwide
Book a demo for the full model list and pricing.
The flagship models enterprises use most
| Model | Vendor | Model ID | Best for |
|---|---|---|---|
| Claude Opus 5.5 | Anthropic | claude-opus-5-5 | Complex reasoning, system architecture, deep code refactoring |
| Claude Sonnet 5.5 | Anthropic | claude-sonnet-5-5 | Code generation, PR review, agent workflows |
| GPT-6 | OpenAI | gpt-6 | Contract review, complex analysis, decision support |
| GPT-5.6 Sol | OpenAI | gpt-5.6-sol | High-volume support, ticket triage, batch processing |
| Gemini 4 Argon | gemini-4-argon | Long documents, research and earnings analysis, multimodal |
Integrate in three steps
If your app already uses the OpenAI SDK, LangChain or Dify, just change the endpoint and key.
Get a department key
An IT administrator issues a Virtual Key for the department in the console and sets its allowed models and budget.
Change the endpoint
Set the base URL to https://api.unirouterai.com/v1 and use a model ID from the table above.
Tag business tasks
Add a task tag to the request header and costs are attributed to that task automatically.
Cost attribution parameters
Two optional headers record calls at a finer level. Without them, cost is billed to the department that owns the key.
- x-unirouter-dept
Overrides the key's default department or team — useful when several teams share one service. Example:
rd-platform - x-unirouter-task
Business task tag, used to count calls and compute cost per task. Example:
contract-review