OpenRouter is the best-known AI model router, and many developers search for how its pricing works. Here is a plain-English explanation, plus how Solo Router does it differently.
How OpenRouter pricing works
At the time of writing, OpenRouter:
- shows each model's price per million input and output tokens, usually matching the underlying provider's price;
- is prepaid: you buy credits and usage is deducted from them;
- charges a fee when you buy credits (a percentage of the purchase), rather than marking up each token.
Prices and fees change, so always check OpenRouter's own pricing page for the current numbers.
What to watch for with any router
- Different providers, different prices. The same model can be served by several providers at different prices and speeds.
- Fees at top-up. A percentage fee on credit purchases adds to your real cost per token.
- Data policy per provider. Check whether the provider serving your request may keep or train on prompts.
How Solo Router prices models
- One price per model, shown per million tokens on the Models page. It already includes our margin.
- Prepaid credit from $5, plus a 5% card processing fee (minimum $0.50) shown before you pay. Per-token prices include our margin, so they are usually a bit higher than buying directly from a provider.
- The exact cost of every request is returned in the API response as
usage.cost. - We do not store prompts or completions, only token counts and cost for your bill. (OpenRouter also does not log prompts by default.)
| Solo Router | OpenRouter | |
|---|---|---|
| Catalog | 25 chat models, open-model focus | Hundreds, including Claude and GPT |
| Billing | Prepaid; one all-in price per model + 5% card fee | Prepaid; provider price + credit purchase fee |
| Per-request cost in response | Yes (usage.cost) |
Via API/dashboard |
If you need Claude or GPT, OpenRouter is the right tool. If you mainly use DeepSeek, GLM, Kimi, Qwen or Llama and want simple billing, try Solo Router. Switching is one line: change the base URL to https://solorouter.ai/api/v1.