Powered by Soterra Labs AI
🔒
Access Restricted
This tool requires a valid access link.
Soterra Labs
Find a well-matched GPU for your workload
Answer a few questions about what you're building and see a GPU recommendation — with pricing and next steps.
5 minutes ~8 questions Personalized recommendation
Powered by Soterra Labs AI
GTM Intelligence
GPU Navigator
Find a well-matched GPU for your workload
GPU Navigator

Find a well-matched GPU for your workload — and estimate your cost.

4 steps, under 5 minutes. See your GPU recommendation, cost estimate, and fit assessment — key factors before your first conversation.

Please enter your company name to continue.
Step 1 of 4 · Workload

1What are you building?
Determines which GPU families are compatible with your workload
Please select a workload type to continue.
2Scale
Narrows the GPU recommendation to a matching performance tier
Please select a scale to continue.
3Check your VRAM fit
Your model size, precision, and context window determine how much GPU memory you need
VRAM = Exact Model Size × Inference Precision × Context Window
3aExact model size
Parameter count for VRAM estimation
billion parameters (B)
3bTraining approach
Determines memory per parameter — lower precision = less VRAM needed
3cContext window · advanced · default 8K
Longer context windows add KV cache memory overhead — only change if your use case needs 32K+ tokens
VRAM Required
4
Different latency needs require different GPU bandwidth — batch vs real-time changes the recommendation
Please select a latency requirement to continue.
5Your recommended GPU is pre-selected — add up to 2 others to compare cost
Already have other GPUs in your fleet? Select them here to see a side-by-side cost comparison
All GPUs for this workload are currently out of stock — contact us to check availability.
Estimated cost comparison
6Estimated GPU hours per month?
Your volume drives the monthly cost estimate and 3-year TCO projection
Please select your monthly GPU hours to continue.
Step 2 of 4 · Current Setup

The tool compares your current provider's rate against the vendor's to estimate potential savings
Please select your current provider.
On-demand vs reserved pricing changes your baseline cost for the savings calculation
Please select your pricing model.
Step 3 of 4 · Requirements

Infrastructure
Hard residency requirements narrow which providers can serve you
Please select your data residency requirement.
Compliance & Contracts
The tool checks whether the provider holds the certifications you need — gaps show up in your report
Please select your compliance requirement.
Active contracts may affect your migration timeline and options
Please select your contract situation.
Step 4 of 4 · Budget & Timing

Helps us tailor the recommendation to your deployment window
Please select your timeline.
Helps us recommend a suitable tier — bare metal GPU typically starts at $5K+/mo
Please select your budget range.
Helps us tailor the next steps to your role in the decision
Please select your decision authority.
If you're open to flexible terms, we can include options in your proposal
Please select your payment preference.