On-demand
Cut your AI bill without cutting quality
Watch the full session on how we cut our own AI bill in half with open-source models, then do the same in your own org. Drop your work email and we will send the recording over.
- How we cut our AI bill 52% with nobody noticing
- A live blind test of open-source vs frontier models
- A cheat-sheet of which model to use for what
Hosted by
Max Brodeur-UrbasCo-founder and CEO
GonzaloChief Product Officer
Models covered
GLMMinimaxKimiDeepSeekOpusSonnetGPT
Get the recording
We will email you the full webinar recording, the same one attendees got.
“Gumloop turned our sales and marketing teams into builders, and saved us thousands of hours of manual work.”
Same output, half the bill
Real numbers from the session, measured on the exact same agents and prompts with no drop in quality.
52%cut from our most-used internal agent. The team could not tell.
85%cheaper for a sales dashboard: GLM vs Opus.
70%cheaper for a customer-feedback deck: DeepSeek vs Sonnet 4.6.
60%cheaper for a sales-enablement deck: Minimax vs GPT 5.5.
Pick the right model
When a frontier model is overkill, and which open-source model to use instead.
The blind taste-test
Frontier vs open-source outputs side by side. See if you can tell which is which.
One-click cost controls
Usage dashboards, org-wide model defaults, and cost simulations built into the product.
Where this is going
Automatic model routing, plus agents that reflect on past runs to cut redundant tool calls.
What you will learn
- Spot where you are overpaying by using frontier models for simple tasks.
- Choose the right open-source model for each job instead of guessing.
- Switch models without losing quality, and how to QA that you did not.
- Cut spend across a whole team with built-in dashboards and one-click defaults.
- Answer the data-privacy and self-hosting questions before they come up.
Want the recording?
Drop your work email and we will send it over.