Pricing
Pay per token. No seat fee.
One balance covers every model on the gateway. Usage is metered on the response, so the number in the console is the number on the invoice.
| Plan | Monthly | Included | Rate limit |
|---|---|---|---|
| Pay as you go | $0 | Metered per 1M tokens | 60 req/min |
| Pro | $49 | $49 of credit, priority routing | 600 req/min |
| Scale | $299 | $349 of credit, dedicated carriers | 3,000 req/min |
Model Square
Three routes live today. Every one of them answers on /v1/chat/completions with the same request shape.
| Model | Vendor | Context | Input / 1M | Output / 1M |
|---|---|---|---|---|
| deepseek-ai/DeepSeek-V4-Flash-0731 | DeepSeek | 128k | $0.14 | $0.28 |
| zai-org/GLM-5.3-Flash | Z.ai | 128k | $0.10 | $0.20 |
| MiniMaxAI/MiniMax-M2.7 | MiniMax | 192k | $0.18 | $0.36 |
# what the model list returns curl https://lazarus-th.duckdns.org/v1/models \ -H "Authorization: Bearer sk-..."
What is not charged.
01
No seat fee
Users are free. You pay for tokens, not for people.
02
No egress fee
Streaming, retries and failed upstream calls cost nothing.
03
No minimum
Top up $5 or $5,000. The balance does not expire.