Got the code? Welcome, builder!
You're heading into Invent the Future — instructors hand out the access code on day one.
You're heading into Invent the Future — instructors hand out the access code on day one.
You've been running prompts across three models. This session is about making a deliberate choice, not just using whatever loads first.
These are general tendencies as of mid-2026. They shift with model updates, and your own experience may differ.
| Model | Strength | Watch for |
|---|---|---|
| Claude (claude.ai) | Follows structured output instructions precisely; detailed CoT reasoning | Can be verbose |
| Gemini (gemini.google.com) | Strong JSON output; Canvas renders HTML live; fast | — |
| ChatGPT (chat.openai.com) | Strong general performance | Tends to wrap output in markdown fences even when told not to |
These are tendencies, not rules. The comparison you're about to run will produce your own observations, which matter more.
All three platforms have free-tier rate limits. On a busy day you will hit them. The best model for a task that rate-limits you mid-build is not actually the best model in practice — the one that's available when you need it is.
That's why you're picking both a primary tool and a backup.
Write a single complex prompt that uses at least three of the five techniques you've learned (zero-shot, one-shot, few-shot, structured output, chain-of-thought). Make it something you'd actually want to build for your Week 2 project.
Run it on all three models. For each run, record:
After running the comparison, write down:
Primary model for Week 2: ___ because ___.
Backup model: ___ (for rate limit situations).
Neither choice is permanent. Switch whenever it's useful. The point of this session is to go into the hackathon with a reasoned starting point rather than defaulting to whichever one you heard about first.