Unified endpoint
One OpenAI-compatible URL for every configured model. Point your existing SDK at it and keep shipping.
One endpoint · configured frontier models
The right model,
at a better rate.
Wholerouter is a single OpenAI-compatible endpoint for the configured Anthropic, DeepSeek, and GLM models your organization can use.
multiple frontier models→1 endpoint
Model roster
Every configured model, one key.
Provider-backed models are reachable through the same endpoint, with a clear published roster.
GLM 5GLM 5 TurboGLM 5.1GLM 5.2
Claude Opus 5Claude Sonnet 5Claude Fable 5Claude Opus 4.8Claude Opus 4.7Claude Opus 4.6Claude Sonnet 4.6Claude Haiku 4.5
DeepSeek V4 FlashDeepSeek V4 ProRates
Transparent rates, built in.
USD per 1M tokens. Rates below are maintained with the published model configuration.
| Model | Input / 1M | Cached input / 1M | Output / 1M | Discount |
|---|---|---|---|---|
| GLM 5Z.ai | $0.750 | $0.150 | $2.50 | 25% off |
| GLM 5 TurboZ.ai | $0.920 | $0.180 | $3.00 | 23% off |
| GLM 5.1Z.ai | $1.00 | $0.800 | $3.30 | 29% off |
| GLM 5.2Z.ai | $1.05 | $0.195 | $3.30 | 25% off |
| Claude Opus 5Anthropic | $4.25 | $0.425 | $21.25 | 15% off |
| Claude Sonnet 5Anthropic | $1.70 | $0.170 | $8.50 | 15% off |
| Claude Fable 5Anthropic | $8.50 | $0.850 | $42.50 | 15% off |
| Claude Opus 4.8Anthropic | $4.25 | $0.425 | $21.25 | 15% off |
| Claude Opus 4.7Anthropic | $4.25 | $0.425 | $21.25 | 15% off |
| Claude Opus 4.6Anthropic | $4.25 | $0.425 | $21.25 | 15% off |
| Claude Sonnet 4.6Anthropic | $2.55 | $0.255 | $12.75 | 15% off |
| Claude Haiku 4.5Anthropic | $0.850 | $0.085 | $4.25 | 15% off |
| DeepSeek V4 FlashDeepSeek | $0.105 | $0.0021 | $0.210 | 25% off |
| DeepSeek V4 ProDeepSeek | $0.326 | $0.0027 | $0.652 | 25% off |
How it routes
Route once. Reach everything.
Explicit provider behavior
WholeRouter keeps routing and spend clear: automatic retries and hidden provider fallbacks are disabled.
One bill, transparent rates
Every request is logged against your organization with published configured token prices.
Getting started
Three steps to your first request.
Sign in to WholeRouter
Start with the partner login flow and access the organization your administrator has provisioned.
Create a virtual key
Use the dashboard to create the LiteLLM virtual key that authorizes your configured model access.
Point your client at one URL
Keep your OpenAI-compatible request shape and send traffic through the WholeRouter endpoint.
FAQ
Clear before you route.
Which models can I use?
The roster is generated from WholeRouter's published configuration and stays in sync with the proxy.
Is the endpoint OpenAI-compatible?
Yes. Use the OpenAI-compatible chat-completions and Responses request shapes with your WholeRouter virtual key.
What is cached input pricing?
Cached input pricing applies only when the upstream provider reports cache-read token usage for a request.
Does Wholerouter automatically fail over requests?
No. Automatic retries and provider fallbacks are disabled so provider behavior and spend remain explicit.
How do I get access?
Partner access starts with the WholeRouter sign-in flow. Your organization then uses LiteLLM virtual keys for inference.
Ready when you are