Gapminder AI Worldview: leaderboard
Gapminder's AI Worldview benchmark — global-development fact questions in the Factfulness tradition, testing calibrated knowledge about the state of the world.
Metric: Correct Rate (%). Source: www.gapminder.org. Saturation forecast: Estimated already saturated. 37 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O3 (Medium) | 93.9 | #124 (O3) |
| 2 | Gemini 2.5 Pro (Preview) | 93.1 | #130 |
| 3 | O1 (Medium) | 90.9 | #183 (O1) |
| 4 | GPT-4.1 | 89.1 | #248 |
| 5 | Claude 3.5 Sonnet | 88.6 | #350 |
| 6 | Claude 3.7 Sonnet (Thinking) | 87.9 | #249 (Claude 3.7 Sonnet) |
| 7 | DeepSeek R1 | 87 | #251 |
| 8 | GPT-4o | 85 | #343 |
| 9 | Claude Sonnet 4 (Thinking) | 84.6 | #199 (Claude Sonnet 4) |
| 10 | Grok 3 | 82.3 | #304 |
| 11 | GPT-4 Turbo | 81.5 | #404 |
| 12 | Gemini 1.5 Pro (002) | 80.9 | #349 |
| 13 | Sonar | 78.4 | #542 |
| 14 | Llama 4 Maverick Instruct | 78.3 | #456 |
| 15 | DeepSeek V3 | 78.1 | #321 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=gapminder-ai-worldview · How It Works · Data refreshed daily, snapshot 2026-10-20.