GPT API
GPT API,
75% below list price.
One API key for every GPT model we offer, billed per token, with OpenAI's list price shown next to each of our prices. Prepay by card, Alipay or WeChat Pay. No monthly fee, no minimum spend.
Save up to 75%vs. OpenAI's list price
Join our Telegram group and talk to us directly if you get stuck.Join the group
Questions, or want to talk volume pricing? Email [email protected]
Price list
USD / 1M tokens · input / output
| Model · 24h success | Our price / List price | Below list |
|---|---|---|
| GPT 6.2 Astra100% | $2.00 / $10.00 | Save 75% |
| GPT 6.2 Sol100% | $0.50 / $2.50 | Save 75% |
| GPT 6.1 Sol98.92% | $0.50 / $2.50 | Save 75% |
| GPT 6.1 Terra99.64% | $0.25 / $1.25 | Save 75% |
| GPT 6 Astra98.34% | $2.50 / $12.50 | Save 75% |
| GPT 6 Luna89.89% | $0.08 / $0.40 | Save 20% |
| GPT 6 Sol96.2% | $0.50 / $2.50 | Save 75% |
| GPT 5.6 Luna100% | $0.16 / $0.96 | Save 20% |
Status
Real success rates, last 7 days.
Each GPT model's request success rate over the past 7 days, one bar per hour, measured on real traffic. Hours without requests are grey.
GPT models
Last 7 days · 1 bar = 1 hourLast 48 hours · 1 bar = 1 hour
- GPT 6.2 Astra100% success
- GPT 6.2 Sol100% success
- GPT 6.1 Sol99.73% success
- GPT 6.1 Terra99.81% success
- GPT 6 Astra98.44% success
- GPT 6 Luna94.79% success
- GPT 6 Sol97.86% success
- GPT 5.6 Luna100% success
Why FlyRouter
Pay as you go. Pay for what you use.
No usage commitment and no subscription. After you top up, every token is charged from your balance at its listed price, so you can start small.
Per-token pricing, in the open
Every model on the price list shows our price next to OpenAI's list price. Input, output and cache tokens each have their own rate.
No usage commitment
No subscription, no monthly fee and no minimum spend. Your balance never expires, and each call is charged from it.
Volume pricing on request
For steady high volume, contact us for a quote based on your usage.
Works with Codex and the OpenAI SDK
Call /v1/chat/completions or /v1/responses with the OpenAI SDK. Codex only needs the endpoint and an API key in config.toml.
One key per project
Each key can call every model, or only the models you pick. A deleted key stops working immediately.
Every call itemized
The usage log lists each call's model, tokens and charge, so you can check them line by line. Calls that fail because of an upstream outage are not charged.
How it works
Your first call in four steps.
All you need is an email address and a way to pay.
01
Create an account
Sign up with your email address.
02
Top up
Pay by card, Alipay or WeChat Pay. Your balance is credited immediately.
03
Create an API key
Choose the models it can call, then copy it somewhere safe.
04
Change the endpoint
Point the OpenAI SDK or Codex at https://api.flyrouter.ai/v1. The rest of your code stays the same.
Referrals
Refer developers,
earn from their top-ups.
Share your referral link. When someone signs up through it and tops up, you earn a share of their top-up, with no cap.
Rebates settle daily and can be withdrawn or moved to your balance for API calls. Rates and rules are on the referrals page.
Get your referral linkFrom the first token,
this is your price.
Start with a small top-up. Prices are listed per million tokens, and every call shows what it actually cost.