Maintain and version data models
The outcome: releases model changes with reasons, impact, semantic versions, migrations, baselines and approval.
Builds on: Generate and review DDL, Review the data model against the scorecard.
Which AI model is best at maintain and version data models?
| Model | Provider | Tier | Test score | Knobits mastered (expected) | Cost per test run |
|---|---|---|---|---|---|
| gemini-4-argon | Flagship | – | 88% | $0.021 | |
| gemini-3.5-flash-lite | Low cost | – | 76% | $0.003 | |
| gpt-6-luna | OpenAI | Low cost | – | 87% | $0.002 |
| claude-haiku-5-5 | Anthropic | Low cost | – | 87% | $0.006 |
Test score: mean rubric score on the test tasks for this activity. Knobits mastered: expected share of the activity's knobits the model masters. Cost: mean API cost of one test run at list prices, half price for answers sent as a batch. Method and caveats.
The 7 knobits of this activity
2 that AI does well, 1 it does partly, and 4 where people still lead.
| ID | Knobit | Type | AI exposure (models mastering) |
|---|---|---|---|
| K-1201 | Recognises a model change with no reason | Skill | People still lead 0/4 |
| K-1202 | Recognises a model change with no impact analysis | Skill | AI does this well 3/4 |
| K-1203 | Recognises a breaking model change released as a minor version | Skill | People still lead 1/4 |
| K-1204 | Recognises a change to a deployed structure with no migration | Skill | People still lead 0/4 |
| K-1205 | Recognises a released model version with no baseline | Skill | AI does this partly 2/4 |
| K-1206 | Recognises a model change released without approval | Skill | People still lead 1/4 |
| K-1231 | Releases a model change as a new model version | Skill | AI does this well 3/4 |