Artificial intelligenceUnified and under control
Hatra gives your company access to AI models, user and usage controls, and AI services. Reduce costs, improve security, and keep control.
OutputsHatraInputs
HHatra
gateway · 0.9 ms p50
Humans
Chat and copilots
Agents
tool loops
Machines
batch jobs
OpenAI
provider keys
Anthropic
provider keys
glm-5.3
onprem-node-01
Open models
catalogue
Your GPUs
owned hardware
AI providers
OpenAI
Anthropic
Google AI
Meta AI
Mistral AI
DeepSeek
Alibaba Cloud
xAI
Cohere
PerplexityOur AI models
gpt-5claude-opus-4-8claude-sonnet-4-5gemini-2.5-prollama-4-maverickglm-5.3qwen3-72bdeepseek-chatmistral-largebge-m3-embed
The problems we solve
Many organizations adopt AI before they are ready. They face problems that can slow their growth and reduce their productivity.
No central control
Without one management layer, it is hard to control user access, usage, limits, and how people use AI.
Data security
Sending company data to different AI services makes it hard to control where it is processed and how it is used.
Separate services
Different models, providers, and tools spread AI management across separate services and environments.
High costs
Using several AI models and services can increase costs and make them hard to predict and manage.
Why Hatra
Power, security, control, and integration. Turn AI into results for your organization.
Security for company data
Protect sensitive data with access controls, security policies, and models hosted on your chosen infrastructure.
99.9%Security and compliance
End-to-end encryptionAdvanced access
Company standards
Central control and governance
Manage and monitor all requests, models, and user access from one place.
Full accessSystem admin
Budget limit activeOrganization user
Team usage limitData analyst
Temporary accessGuest
Integration and deployment
Connect to your existing infrastructure and deploy at scale with full integration.
SAPMicrosoft 365
API GatewaySentrySlackOkta SSOSCIM
Reliability and performance
Use routing, fallback, and smart caching for stable responses and better performance.
Under 1 secondRoute to cache
Server errorSwitch to another model
Under 50 msLoad from memory
Budget limitBlock the request
Run on your own GPUs
Serve open models on your hardware or Hatra cloud infrastructure at a lower cost.
ActiveGPU-01 processing server
$3,410 savedThis month
GPU use78%
Control and reduce AI costs
Manage team and model usage to make AI costs more predictable and efficient.
Data Science
Engineering
Marketing
Sales
▼ 40%Using open models
One dashboard for your organization's AI
View and manage all AI models, resource use, costs, and performance in one place. Keep control without switching between tools.
console.hatra.io/costs
Clear budgets, controlled spending
Set budgets for each key and team. Get alerts before reaching a limit.
Credit remaining
8,400,000
toman
Spent this month
3,600,000
30% of the monthly budget
Monthly limit
12,000,000
toman for this workspace
Team budgets
1,200,000 of 4,000,000 toman
Product team
1,800,000 of 3,000,000 toman
Engineering team
600,000 of 5,000,000 toman
Support team
Usage limits are active. Requests over budget stop.
Feature preview. Sample data.
AI within your organization
Run models, data, and the management console on your own infrastructure to meet your security and architecture needs.
Control the data path
Choose which models and infrastructure handle your data. Keep sensitive data away from external providers.
Run models on your own servers
Run internal models on your own servers so data processing stays within your organization.
Dedicated models hosted by Hatra
Run the models your organization needs on dedicated Hatra infrastructure, without relying on external providers.
Host all of Hatra on your servers
Host the Hatra console and services within your organization. Use them without an internet connection.
GPU infrastructure for AI
Activate the GPU resources you need and manage your running infrastructure from the Hatra panel.
Run your own models
Run your AI models on dedicated GPUs. Keep resources private and performance stable.
Monitor and manage from your panel
Monitor GPU status, resource use, and infrastructure health from your Hatra dashboard.
Scale with demand
Increase or reduce GPU resources as request volume changes. Avoid a fixed capacity limit.
console.hatra.io / gpu-clusters
| Status | GPU workload | Hardware model | Server ID |
|---|---|---|---|
| Active | 84% | NVIDIA H100 SXM5 | hatra-node-gpu-01 |
| Active | 62% | NVIDIA H100 SXM5 | hatra-node-gpu-02 |
| Active | 45% | NVIDIA A100 80GB | hatra-node-gpu-03 |
| Active | 12% | NVIDIA L40S | hatra-node-gpu-04 |
| Standby | 0% | NVIDIA H100 SXM5 | hatra-node-gpu-05 |
hatra-node-gpu-01
NVIDIA H100 SXM5Hardware model
84%
Activehatra-node-gpu-02
NVIDIA H100 SXM5Hardware model
62%
Activehatra-node-gpu-03
NVIDIA A100 80GBHardware model
45%
Activehatra-node-gpu-04
NVIDIA L40SHardware model
12%
Activehatra-node-gpu-05
NVIDIA H100 SXM5Hardware model
0%
StandbySubscription plans
A 30 minute demo with your own data. We connect a provider key and show your first report before the call ends.
All features
Enterprise
Manage your team with access to different models, a monitoring dashboard, and usage reports.
- All Standard plan features
- Dedicated support
- Your own URL
- Guaranteed SLA
- Custom features for your needs
Free
Standard
Manage your team with access to different models, a monitoring dashboard, and usage reports.
- Team and access management
- Add and use different models
- Monitoring dashboard
- Usage and cost reports
- Enterprise security
We had four teams, three provider accounts, and no idea who was spending what. A week after we put Hatra in front of it, I could answer the question of the CFO on one screen. And the budgets stop the spend instead of reporting it.
Marcus Vaillancourt
Platform Lead, Cobalt Systems
- Subscription plan
- Pro
- Service satisfaction
- 98%
- Completed projects
- 15
- Time using the service
- 2 years
FAQ
Questions we get asked
Hatra is an AI Gateway. It gives organizations and developers safe, central access to multiple local and international AI models at the same time.
Everything you need to run AI in your organization
From $12 per person a month, or 4% of model spend. Connect one provider key and see the first report before the call ends.
