Which LLM should I use?
Answer five questions about your task, priorities and constraints. The picker scores every current model in our directory and returns three to test, with the reasons for each.
From shortlist to decision
The picker uses what providers publish: positioning, prices, context limits and input types. It can't see how a model handles your documents, your codebase or your customers. Close the gap with a quick test.
- Collect 20-50 real examples of the task, including a few hard ones.
- Write one prompt and run it unchanged on all three models.
- Score the outputs against what a good answer looks like. A simple pass or fail per example is enough to start.
- Divide each model's cost by its passes. The lowest cost per correct answer usually wins, unless latency rules it out.
The full framework, with a weighted scoring matrix and a procurement checklist, is in how to choose an LLM.
Still choosing a model?
Send us your use case and constraints. We'll come back with a shortlist and an evaluation plan you can run on your own data.
- Free and vendor-neutral: no model provider pays for placement.
- Your details go only to partners you agree to.
Frequently asked questions
How does the model picker choose?
It removes models that can't meet your hard requirements (self-hosting, image or audio input, context length), then scores the rest on how well their provider's stated strengths match your use case, their capability tier, their blended price and a speed proxy. Your priority answer changes how much each factor counts.
Should I trust the top pick blindly?
No. The picker narrows dozens of models to three plausible candidates using published facts. The best model for you is the one that does your actual tasks well at an acceptable cost, so run a small test with real examples before deciding.
What if my use case isn't listed?
Choose the closest one, or pick general assistant. You can also describe your project on our expert help page and get a vendor-neutral shortlist from a person.