GPT API pricing, at a fraction of the list.
The GPT models cost 30% of OpenAI's official API price through Chernion, billed per token from a prepaid balance. Keep the OpenAI SDK you already use. The only change is the base URL and the key.
Per million tokens, official struck through.
The faint figure is OpenAI's official list price. The bold figure is what you pay through Chernion, which is 30% of official for the GPT group. Rates are read live, so this table is always current.
| Model | Input / 1M | Output / 1M | Save |
|---|---|---|---|
GPT-5.4 gpt-5.4 | $2.50$0.750 | $15.00$4.50 | −70% |
GPT-5.4 mini gpt-5.4-mini | $0.750$0.225 | $4.50$1.35 | −70% |
The frontier GPT and coding models sit in the premium group at 50% off on the full model list.
A fixed share of official, billed per token.
Every GPT model is billed at a fixed percentage of OpenAI's official API price, set by its group. The GPT group is 30% of official, so a model that lists at 2.50 per million input tokens costs 0.75 here. You pay per token for the input you send and the output you get back, with no subscription and no minimum.
- 1 million input tokens
- $0.750
- 1 million output tokens
- $4.50
- A 2,000 in and 500 out request on GPT-5.4
- $0.003750
Tokens are the unit of text the model reads and writes, roughly four characters each. You are billed for the input you send and the output you get back, nothing else.
Change two lines, keep your stack.
- Point base URL at the Chernion endpoint and swap in your key. Your existing OpenAI SDK calls keep working.
- Streaming and tool calls are supported exactly as the OpenAI API defines them.
- Every response returns its exact cost in a chernion field, so spend is visible to the token.
- Top up the prepaid balance with crypto or a card, with no subscription and no minimum.