Find my match
Three quick questions. At the end: the model that fits your situation, the minimum honest build to run it, and the total price — with an honesty note if your budget and your ambitions disagree.
1. What are you building?
Just me, or a small team experimentingLearning, prototypes, internal tools
A startup shipping ONE AI productOne big model behind your product, run as cheaply as it honestly can be
A company serving many usersReal traffic today, and it keeps growing
2. What budget feels right?
Under $50,000Workstation-card territory — still enough for a 235B model, honestly
$50,000 – $500,000Datacenter GPU territory — real H200-class capacity
$500,000+Multi-node territory — the biggest open models
← Back
3. How fast will your needs grow?
Just experimenting for nowNo pressure to scale yet
Steady growthMore users over time, nothing explosive
Fast growthWe'll need serious headroom soon
← Back
This wizard needs JavaScript. The same guidance (with the full math) lives at Help me choose .