Simple pricing. No surprises.
Prepaid credits, actual usage metering, and detailed spend reports. No credit purchase fees, no hidden charges, no VC investors to pay back.
- 25+ free models (DeepSeek, Llama 4, Gemini Flash)
- No credit card needed
- Community rate limits
- OpenAI-compatible API
- Waterfall smart routing
Good for: Experimentation, hobby projects, learning
Start Free- All 440+ models (GPT-5, Claude, Gemini Pro, etc.)
- $5 minimum credit purchase
- Detailed spend by strategy, model, and request
- Pay with USDC on Base via x402 - no Stripe
- Card credits via Stripe with a $25 minimum
- Free models still free
Good for: Production apps, indie developers, agents
Get StartedCard credits: $25 min- Custom rates below 4%
- Dedicated support from Julie
- SLA guarantees
- Custom model configurations
- Priority bug fixes
Good for: High-volume users, startups scaling up
Contact JulieWaterfall vs OpenRouter
They raised $40M. We didn't. Guess who's cheaper.
| Feature | OpenRouter | Waterfall |
|---|---|---|
| Credit purchase fee | 5.5% | None |
| Pricing model | Per-token + 5.5% credit fee | Prepaid credits, metered by actual usage |
| Free models | ~20 | 25+ |
| Hidden fees | Credit purchase fee | None |
| Funding | $40M VC-backed | Bootstrapped |
| Team size | 20+ engineers | Julie + Claude |
Strategy Pricing
Pick a routing strategy, pay a flat fee per request. We handle the model selection.
| Strategy | Price / request | What you get |
|---|---|---|
| budget, free_smart | Free | Free models only — 25+ options |
| speed_first | $0.002 | Fastest models (Groq, Cerebras) |
| coding, vision | $0.002 | NVIDIA free + paid fallbacks |
| reasoning | $0.003 | Chain-of-thought (DeepSeek R1 + o3) |
| context_max, multilingual | $0.003 | Large context / polyglot models |
| smart_video | $0.003 | Vision/video — NVIDIA + Gemini |
| privacy_smart | $0.005 | Zero Data Retention — 348+ ZDR models |
| smart_legal, smart_health | $0.005 | Privacy-first NVIDIA + Claude/GPT |
| tool_calling | $0.005 | Reliable function calling |
| smart_image | $0.005 | Image generation — Gemini Flash Image + GPT-5 Image |
| smart_voice | $0.005 | Voice agents — Gemini audio + GPT-4o audio |
| smart_transcribe, smart_multilingual_voice | $0.003 | Audio → text + multilingual voice |
| quality_first, agentic, research | $0.010 | Frontier models (GPT-5, Claude Opus, o1, Perplexity) |
Frequently asked questions
No sales calls. Just answers.
How does pricing work?
You buy prepaid credits, then each request draws down based on the actual model Waterfall used and the actual tokens reported by the provider. Free models are always free. Paid usage includes a detailed breakdown by strategy, model, date, and time.
Why is it cheaper than OpenRouter?
We're bootstrapped. One person, low overhead, no investors demanding returns. No VC money to burn, so we keep costs low. Your money goes to AI providers, not our office lease.
Is there a minimum spend?
The free tier requires no payment info. Paid credits have a $5 minimum with x402 or a $25 minimum by card. After that, every request draws down based on actual token usage.
Do you support crypto payments?
Yes. USDC on Base via the x402 protocol can be used for credit purchases. We use credits for normal usage because it is safer and gives better usage reporting.
Can I pay by card?
Yes. Card credit purchases run through Stripe with a $25 minimum. x402 stays available for lower-minimum $5 credit purchases.
What about rate limits?
Free tier gets community rate limits -- enough for development and hobby projects. Paid tier gets generous limits based on your spend. Volume customers get custom limits.
Can I use this in production?
Yes. We have uptime monitoring, fallback routing across multiple providers, and automatic failover. If one provider goes down, your requests route to the next one -- that's the waterfall.
Stop overpaying for AI
Same models, lower prices. No credit fees, no hidden charges. Built by one person who actually uses this stuff.