Metered in ciphertext.
Never in content.
We can't read your traffic, so we bill the only things we can see: bytes and requests. Every plan includes the full RFC 9458 relay surface, and our zero-logging guarantee: nothing retained beyond billing records.
For development, testing, and evaluation.
- Shared relay endpoint
- 50K requests / month
- 50 GB bandwidth / month
- 1 MB max request size
- Classic OHTTP (message/ohttp-req)
- HTTP/1.1 + HTTP/2
- Fair-use rate limits (2 req/s max)
- Community support
For side projects going live.
- Dedicated relay endpoint
- 5M requests / month
- 150 GB bandwidth / month
- 50 MB max request size
- 25 req/s, 2× regional bursting
- HTTP/1.1 + HTTP/2
- Chunked OHTTP (streaming)
- mTLS to your gateway
- Acceptable Use Policy
- Usage analytics dashboard
- Email support
For services shipping OHTTP to real users.
- Everything in Starter
- Unlimited requests
- 500 GB bandwidth / month
- 100 MB max request size
- 150 req/s, 2× regional bursting
- HTTP/1.1 + HTTP/2 + HTTP/3
- Auto provider compliance
- Privacy Pass support
- Configurable rate limits
- 99.9% uptime SLA
- Priority support
For high-volume, latency-sensitive fleets.
- Everything in Scaling
- Unlimited requests
- 4 TB bandwidth / month
- Unlimited request size (fair use)
- 1,000 req/s, 2× regional bursting
- HTTP/1.1 + HTTP/2 + HTTP/3
- Privacy Pass issuance
- Auto provider compliance
- Region pinning
- Gateway-in-a-box templates
- 99.95% uptime SLA
- Dedicated support channel
For regulated industries and dedicated infrastructure.
- Everything in Pro
- Custom request & bandwidth limits
- Custom request size & rate limits
- Single-tenant relay deployment
- Custom regions & pinning
- Custom SLA & support contract
- Security review & onboarding
- Invoiced billing
Operating an Oblivious Gateway at scale? The Gateway Partner program offers single-tenant relay deployments, joint key-rotation runbooks, and traffic-analysis resistance reviews.
Compare the tiers.
Every tier includes the full RFC 9458 relay surface, zero logging beyond billing records, and our multi-provider anycast network. Scaling and Pro bill on bandwidth moved, with unlimited requests.
| Community | Starter | Scaling | Pro | Enterprise | |
|---|---|---|---|---|---|
| Price | $0 | $20 / mo | $49 / mo | $399 / mo | Custom |
| Requests / month | 50K | 5M | Unlimited | Unlimited | Custom |
| Bandwidth / month | 50 GB | 150 GB | 500 GB | 4 TB | Custom |
| Max request size | 1 MB | 50 MB | 100 MB | Unlimited (fair use) | Custom |
| Max request rate | 2 req/s | 25 req/s | 150 req/s | 1,000 req/s | Custom |
| Regional bursting | ✗ | 2× regional burst | 2× regional burst | 2× regional burst | Custom |
| Relay endpoint | Shared | Dedicated | Dedicated | Dedicated + capacity | Single-tenant |
| Classic OHTTP (message/ohttp-req) | ✓ | ✓ | ✓ | ✓ | ✓ |
| Chunked OHTTP (streaming) | ✗ | ✓ | ✓ | ✓ | ✓ |
| HTTP/1.1 + HTTP/2 | ✓ | ✓ | ✓ | ✓ | ✓ |
| HTTP/3 (QUIC) | ✗ | ✗ | ✓ | ✓ | ✓ |
| mTLS to your gateway | ✗ | ✓ | ✓ | ✓ | ✓ |
| Acceptable Use Policy | Fair-use | ✓ | ✓ | ✓ | ✓ |
| Auto provider compliance | ✗ | ✗ | ✓ | ✓ | ✓ |
| Privacy Pass support | ✗ | ✗ | ✓ | ✓ | ✓ |
| Privacy Pass issuance | ✗ | ✗ | ✗ | ✓ | ✓ |
| Usage analytics dashboard | ✗ | ✓ | ✓ | ✓ | ✓ |
| Configurable rate limits | ✗ | ✗ | ✓ | ✓ | ✓ |
| Region pinning | ✗ | ✗ | ✗ | ✓ | ✓ |
| Gateway-in-a-box templates | ✗ | ✗ | ✗ | ✓ | ✓ |
| Uptime SLA | ✗ | 99.9% | 99.9% | 99.95% | Custom |
| Support | Community | Priority | Dedicated channel | Dedicated team |
For reference: comparable hosted OHTTP relays start at $15/month for 2.6M requests (~$5.73 per million) and offer 13.1M requests at $50/month (~$3.81 per million).
Auto provider compliance. The unlinkability guarantee quietly breaks if your relay and gateway end up on the same network. If you spin up a gateway on Fly.io, AWS, or any provider whose infrastructure we also run on, the relay can see it is talking to the same operator, and the privacy guarantee erodes without anyone noticing. On Scaling and Pro we continuously check where your configured gateway actually resolves and routes, verify it sits on a network independent from ours, and alert you - or refuse the mapping - the moment that stops being true. No other hosted relay does this.
Questions, answered.
Is the free tier really free?
Yes. Community is a shared, fair-use relay with 100K requests a month for development, testing, and evaluation. No credit card required, unlike other hosted relays, we don't charge you to evaluate the protocol.
How is usage measured if you can't read requests?
By ciphertext bytes and request counts only. Billing runs on volume and timing metadata, never on encapsulated content, which we cannot see.
Can ohttp.io relay to any gateway?
Yes. Each relay endpoint maps one-to-one to a gateway URI you configure, as RFC 9458 requires. The gateway can be yours or a partner's.
What is regional bursting?
Paid plans can burst to 2× their stated request rate per region for short spikes, so a launch-day surge in one geography doesn't trip the limiter while the rest of your traffic stays at its normal rate. The Community tier has a fixed 2 req/s ceiling with no bursting.
Which plans include chunked OHTTP, mTLS, and Privacy Pass?
Chunked OHTTP and mTLS to your gateway start at the Starter tier ($20/month). Privacy Pass support is included from Scaling ($49/month), and full Privacy Pass issuance, where we operate token issuance for you, is part of the Pro tier.
What happens when I exceed my plan's allowance?
On the request-capped plans (Community, Starter) we throttle gracefully with 429 responses when you hit the monthly request ceiling, and notify you before limits bite. On the unlimited-request plans (Scaling, Pro) bandwidth overage past your allowance is billed at $0.05/GB. Upgrades apply instantly.
Is there an SLA?
Yes, 99.9% uptime on Starter and Scaling, 99.95% on Pro, backed by service credits. Enterprise SLAs are custom. Live and historical uptime is public at status.ohttp.io.