Generative AI Model Selection
Choose a model using a task-specific evaluation rather than a popularity claim.
How it works
A useful choice considers task quality, allowed data handling, latency, cost, modalities and operational limits. Compare candidates on the same representative inputs and scoring rules. Record failures as well as successes. Older names such as GPT-3.5, GPT-4 and Bard appear in historical material; do not assume an old model or interface remains available merely because an old tutorial uses it.
How to use this lesson
This is an authored prompt exercise with a human-written acceptance guide. It is not a captured provider response or a promise that a model will produce the target. No provider account, API, model weights or live inference was used for the course execution review. If you try it in an external service, use an authorised environment, check its actual access and data settings, and record the product/model and run date.
Bhylu does not run a generative model or send this prompt to an AI service. Copy and download preserve the text only. External tools may require accounts, paid access or compatible inputs; this course does not promise availability or prices.
Prompt exercise
Evaluate a candidate on this task:
Extract the start time from "Lantern frame session starts at 14:00".
Return only {"start_time":"HH:MM"}.
If no start time is supplied, return {"start_time":null}.
Acceptance guide — not a model transcript
The known-time case yields 14:00; the missing-time case must remain null.
Use both ordinary and missing-input cases when comparing candidates. A fast answer that guesses a time fails the contract. Provider pages in this course identify products without promising current prices, free access, exact quotas or a universal best model. Check the linked official information before adopting a service.