Model Match The right AI for the job.
Tell us what you're making and we'll help you choose a language model, image model, and cost-conscious alternative.
Quick Match
If you're doing this... tap it and get a starting point right away.
The three rules of Model Match
Everything the matcher recommends comes from these.
Don't pay for intelligence you don't need.
Simple summarization, brainstorming, categorization, trivia, and short answers often work well with faster models.
Spend more where quality matters.
Complex instructional design, nuanced writing, multi-step reasoning, and difficult workflows may justify a stronger model.
Turn image generation off when you don't need it.
If your application does not create images, leave the image model off.
Bonus rule: test before you pay more.
Start with the recommended model. Try the budget option. If you cannot tell the difference in your actual app, use the cheaper model.
Suggested stacks
Common combinations of a language model and an image model for TeacherHive apps.
Guided Model Matcher
Five quick questions. Switch modes any time.
If you're doing this...
A searchable reference of common tasks, a model to start with, a budget option, and when to upgrade. Tap any row's task to see a full recommendation and example prompt.
| Task | Start with | Why | Budget option | Upgrade when |
|---|
Model Explorer
Every model currently listed in TeacherHive, in plain English. Cost tiers are relative to each other, not exact prices.
Ratings are simplified Model Match guidance for comparing use cases. They are not standardized benchmark scores.
Language models
Image models
Model Test Lab
The cheapest model that does the job well is the right model. Here's how to find out which one that is, using your real prompts.
Testing checklist
Check items off as you go, then copy or print the list for your notes.
What "good enough" looks like
If the lower-cost model meets your accuracy, instruction-following, safety, formatting, and quality requirements, use it. Differences that don't meaningfully improve the user's experience may not justify the added cost. Wrong facts, ignored instructions, unsafe output, and broken formatting always do.
About Model Match
A small guide for a big question: "If I am doing this, which model should I use?"
What this is
TeacherHive creators often have access to several AI providers and models, but it isn't obvious which one fits a given task. Model Match gives practical starting points based on the kind of work you're doing, how complex it is, what you care about most, how often it will run, and whether images are involved.
It does not rank companies or declare a universal winner. The recommendations are starting points, not verdicts, and the right answer for your app is whatever performs well when you test it.
How the recommendation works
Model Match makes no AI requests. Your Model Matcher selections are evaluated locally in your browser by a simple rules engine and are not sent to an AI model. Each model in the catalog has plain ratings for quality, speed, cost, complexity, volume suitability, structured output, writing, and reasoning. Each task type has weights for which of those matter. Your answers about complexity, priority, and usage volume adjust the weights, the scores are added up, and the best fit, an alternative, and a budget option come out the other end.
If you can read a spreadsheet, you can read the rules. Look for MODEL_CATALOG, TASK_TYPES, and TASK_EXAMPLES in the page source. Adding or changing a model means editing one entry there, nothing else.
What you should leave with
- The strongest model is not always necessary.
- High-volume apps should consider cost and speed.
- Complex tasks may justify premium models.
- Image generation should only be enabled when needed.
- Models should be tested with the actual application before deployment.