Pay per token. Or subscribe.
Reseller rates for 18 frontier models — cheaper than buying direct, with one API key. Pay-as-you-go per 1M tokens, or pick a daily-quota Sub (free models included).
Free tier
Qwen 3.7 is $0.00 on input and output — 1M context, 130 tok/s. No credit card, no expiry.
Pay-as-you-go
Billed strictly per token with no seats, no minimums and no expiry on credit. Stop whenever you want.
Volume discounts
Past $500 of monthly spend, rates drop automatically — no negotiation, no contract. Contact sales for committed use.
Huge daily quotas. Below-market price.
Free models are unlimited (fair use). Paid models share your plan’s daily quota; premium models simply include a smaller token allowance, shown on each row. Quotas reset every day — no top-ups, no per-request credit packs.
Starter
$4.99/mo
1.5M quota tokens/day
≈ 45M quota tokens / month
- ✓ +$3 bonus credit on activation
- ✓ Free models unlimited (fair use)
- ✓ 20 TTS requests/day
- ✓ OpenAI-compatible API key
Pro
$12.99/mo
6M quota tokens/day
≈ 180M quota tokens / month
- ✓ +$5 bonus credit on activation
- ✓ Free models unlimited (fair use)
- ✓ 60 TTS requests/day
- ✓ Priority routing + email support
Ultra
$29.99/mo
20M quota tokens/day
≈ 600M quota tokens / month
- ✓ +$12 bonus credit on activation
- ✓ Free models unlimited (fair use)
- ✓ 150 TTS requests/day
- ✓ Priority routing + dedicated support
Your quota is shared by model cost. On Pro, the daily allowance is about 3M GLM Flash, 750K GPT-6, or 18.8K Claude Opus tokens; Starter and Ultra scale with the plan. Quota resets daily; no resale.
Full price list
Output speed measured as median over the last 24 hours.
| Model | Context | Output speed | Media | Input / 1M | Output / 1M | Actions |
|---|
Prices are per 1M tokens on the standard tier. Failed requests (5xx from our side) are never billed. Latency measured from us-east-1.
Pricing FAQ
How is billing calculated exactly?
Input and output tokens are metered separately and summed at the rates above. Prompt caching and failed requests are not billed. One balance covers every model.
Do you charge for embeddings or image models?
We don't sell embeddings. Image input is included in the normal per-token price for models marked IMG; video-capable models are marked VID. There is no separate vision surcharge.
Why are your rates lower than the providers'?
We benchmark against OpenRouter and OpenCode's listed rates: Claude Opus/Sonnet, GPT-6, DeepSeek 1017 and GLM Flash are roughly 10–25% below them. Grok and Kimi are 10–15% below, and Gemini 4 Argon is lower still because we route through limited-time partner capacity. The numbers on this page are our actual sell rates — no hidden markup at checkout.
What happens when my balance hits $0?
Paid models return a 402 until you top up; free-tier models like Qwen 3.7 keep working. We email you before the balance runs out if usage alerts are enabled.
Start with $0.00
Create an account, grab an API key and ship your first request today — free models included.