|
AI
| #263
ChatGPT, Claude, Gemini: Which AI model should you use?
Want to see these tips early? Subscribe to the newsletter!
Ok so let me get this straight:
- ChatGPT has: Astra, Sol, Terra, Luna…
- Claude has: Fable, Opus, Sonnet, Haiku…
- Gemini has: Pro, 3.8 Flash, 3.7 Flash… (lol Google)
Not to mention the effort levels: Medium, High, Extra High, Ultra. Ok they’re just messing with us at this point right? Which one should we be using? 😭
Here’s a simple analogy:
- Steve Jobs: What should we build, and why would anyone want it?
- Tim Cook: How do we deliver that consistently to millions of people?
Use Astra/Fable/Pro for the Jobs part: figuring out the approach when the work is difficult and unfamiliar.
Once you have a process you’re happy with, use Sol/Opus/Flash for the Cook part: running it again with this week’s information.
On effort levels, I’m going to oversimplify a bit and just tell you to default to High.
- The labs themselves have shared data showing Extra High can spend more time thinking for many tasks without giving you a better answer.
- Compared with Medium, High provides a meaningful enough increase in output quality that makes the extra thinking time and usage worth it.
Putting this into practice
Imagine you just got promoted (more work with the same pay?! SIGN. ME. UP.) and inherited a product launch project. Sales wants to launch next month, Product says it needs more time, and Support is still waiting for training.
Give Astra, Fable, or Pro their strategy decks, your manager’s brief, and recent meeting transcripts, then ask:
Based on these materials, what are my top 3 priorities? For each priority, develop a repeatable process I can run with AI. Include the inputs, steps, and expected output, and flag anything I should clarify with my manager.
That could lead to 3 processes:
- Keep the launch on schedule: A weekly review that checks each team’s deadlines and flags anything that would delay another team.
- Get Support ready: An FAQ workflow that matches customer questions to product docs and flags the gaps Product needs to fill.
- Get decisions from your manager: A Friday update covering progress, risks, and the specific decisions you need them to make.
Next Friday, give Sol, Opus, or Flash the manager-update instructions and that week’s team notes. If Product’s testing has slipped, it can use those instructions to flag the impact on Sales’ launch date and the decision your manager needs to make.
Basically, use Astra/Fable/Pro to work out a difficult process, then Sol/Opus/Flash to help you run it.
3 ways to use this yourself
- Plan your week: Ask Gemini Pro to review your Gmail, Google Calendar, and project files in Drive, then create rules for prioritizing work around your responsibilities and deadlines. Each Monday, have Gemini Flash use those rules to prioritize new tasks and fit them around your meetings.
- Write recurring emails: Give Fable a few emails you’re happy with and ask it to create a reusable guide to your tone, structure, and what a good reply should cover. Then have Opus follow that guide when drafting replies to new threads.
- Compare software: Give Astra your team’s requirements, budget, and frustrations with your current software, then ask it to build a comparison checklist. Have Sol evaluate each new option against that checklist using its feature and pricing pages.
Whenever you’re ready, here are some other ways I can help you:
🎛️ The Cowork Academy: Ready-to-install AI Command Center that loads your context automatically and makes every Claude session sharper than the last. 90% built when you download it.
💻 The Workspace Academy: Never lose a file, task, or note again with my CORE workflow for Google Workspace. Adopted by over 10,000 Googlers.
✅ Build Your Command Center in Notion: Plug-and-play dashboard that surfaces today’s priorities automatically and organizes your life in seconds. #1 rated Notion Course globally.