🚀 B200 bare metal now at $5.6/hr. The best price you'll find. DC in US West → (Access it from Bare metal button on top after login).

Get Your B200 →
Start Building →
Contact · Got questions?

Talk to a human.

packet.ai is designed to be self-serve. Sign up, deploy a GPU in 5 minutes, no calls required. But if you've got questions about pricing, GPUs, networking, security, or billing, drop us a note. A real engineer will get back to you within 24 hours.

  • Engineers, not SDRs
    Your message goes to people who know CUDA, networking, and the platform.
  • Self-serve first
    Sign up free, no card required. Try it yourself before talking to us.
  • Fast response times
    Most questions answered within hours. P1 support under 15 minutes.
No sales pitch. Promise.
Self-serve from $0Real engineersBuilt for developers
We'll only use your details to follow up on this question. See our Privacy Policy.
Other ways to reach us

One inbox. Real engineers.

Whatever you're writing about, send it to help@packet.ai. Your message goes directly to engineers who know the platform. No chatbots, no ticket queues, no tiered support.

Technical support
CUDA, networking, drivers, performance, billing, integrations.
help@packet.ai
General questions
Product feedback, integrations, partnerships, or just to say hi.
help@packet.ai
Security & compliance
SOC 2 reports, DPAs, vulnerability disclosure, audit support.
help@packet.ai
Volume & enterprise
Multi-node clusters, reserved capacity, custom contracts, MSA reviews.
help@packet.ai
Common questions

Before you write to us…

You might find the answer here. Most things on packet.ai don't need a conversation. The platform is designed to be self-serve. But if you don't see your question, drop us a note above.

How do I get started?
Sign up at packet.ai. No credit card required to start. Deploy a GPU in under 5 minutes via the dashboard, CLI, or REST API. The docs have working code samples for vLLM, PyTorch, and JupyterLab.
Do I need to talk to sales?
No. packet.ai is fully self-serve from $0. Pay by the hour, no minimums, no contracts. You only need to email help@packet.ai if you want multi-node cluster reservations, custom contracts, or an enterprise compliance review.
Which GPUs are available?
NVIDIA B200 (180GB HBM3e), H200 (141GB HBM3e), RTX 6000 Pro (96GB GDDR7), plus H100 SXM, A100 80G, and RTX 5090 for cost-optimized inference. Multi-node InfiniBand clusters available on request.
How does pricing work?
On-demand: H100 $1.95/hr, H200 $2.55/hr, B200 $3.75/hr, RTX 6000 Pro $0.66/hr. Per-hour billing, no minimums. Volume discounts kick in automatically above 1,000 GPU-hours/month.
Do you train models on customer data?
Never. Per our DPA, packet.ai does not access, inspect, or use customer datasets, models, workloads, or outputs for training, fine-tuning, benchmarking, or product development. Read the full DPA →
Are you SOC 2 / GDPR compliant?
SOC 2 Type II report and ISO 27001 alignment available under NDA. Email help@packet.ai. GDPR-compliant with a published DPA and SCCs for international transfers.
What's your response time?
General questions: typically within a few hours during business hours, 24 hours otherwise. P1 production support: under 15 minutes, 24/7. We don't use tiered support queues. You talk to engineers directly.
Where are your GPUs located?
Active capacity in us-east, us-west, and EU regions. Specific data centers shown at provision time. EU customers can pin workloads to EU regions for GDPR alignment.
Headquarters

Built in San Jose, shipping globally.

packet.ai is operated by HOSTED AI Inc., a Delaware corporation headquartered in San Jose, California. Infrastructure capacity across the United States and the European Union.

HOSTED AI Inc.
622 North 9th Street
San Jose, CA 95112, USA
help@packet.ai