Launched this week
PromptUnit
Cut your AI API bill without losing answer quality
4 followers
Cut your AI API bill without losing answer quality
4 followers
PromptUnit checks your app's AI requests side by side and moves each type of request to a cheaper model only when it answers just as well. Connect with two settings, keep your own provider keys. Free until it saves you money, then 20% of what it saves.
Free
Launch Team
Render Sandboxes: a computer for every agentRun untrusted code safely. Idle sandboxes cost nothing
Promoted
Hi Product Hunt! I'm Igal, the founder.
I built PromptUnit after seeing how many AI requests go to the most expensive model by default, even when a cheaper one would give the same answer.
How it works: you point your app at PromptUnit (two settings: the API address and a PromptUnit key). No code? Paste one ready message to your AI builder and it makes the change. Your own OpenAI, Anthropic, Google, Groq, DeepSeek or OpenRouter key stays yours.
For each type of request, PromptUnit compares answers from cheaper models side by side. Only when they're just as good does it move that request type over. If quality drops, it moves back. On OpenRouter or with several keys, it picks across all your models.
You see every request on the dashboard: the model you asked for, the model that answered, the cost and the saving. It's free until it has saved you money, then we take 20% of the saving.
I'd love your feedback, especially from anyone running AI features in production.