Choose an AI provider

I gave Claude Code, Codex, and Posit Assistant the same prompts on an R script and two data files about Oregon food pantries, with errors planted in both. All three handled the task well, so I compared them on cost, model choice, and data retention instead.

You'll learn:

• A way to test AI assistants by planting known errors in a script and its data, then checking whether each one finds them, including a trick question about a drop that isn't real
• How the $20 per month tiers differ: Claude Pro and ChatGPT Plus give a pool of usage, while Posit AI Pass gives $20 of credit and shows the cost of each message
• Why Posit AI Pass lets you switch between models, such as Claude Opus 5.5 and GLM 5.3 Flash, and what a zero retention data agreement means
• How to choose a provider for everyday R work: use what you already pay for, or try Posit AI Pass if you want to compare models

Tools mentioned: Claude Code, Codex, Posit Assistant, Posit AI Pass, Claude Pro, ChatGPT Plus, Claude Opus 5.5, GLM 5.3 Flash

Key moments:

• 00:03 Which AI provider to use for R
• 00:44 Test setup with fake food pantry data and planted errors
• 01:22 First prompt: what does the script do?
• 01:38 Posit Assistant uses other models, switching to GLM 5.3 Flash
• 02:03 Trick question about visits dropping in the fall
• 02:39 Fixing the inflated statewide total
• 03:15 Cost comparison across the three providers
• 04:28 Why Posit AI Pass may be worth it
• 04:54 Zero retention data agreement
• 05:30 Which provider to choose