Grok-1.5V
xAI previews Grok-1.5V, its first multimodal model, and introduces the RealWorldQA spatial-understanding benchmark.
Reads documents, charts, screenshots and photos alongside text. xAI reports 68.7% on its own new RealWorldQA benchmark versus 61.4% for GPT-4V. Announced as coming 'soon' to early testers; no public release followed in the sources checked.
- Date
- Friday, 12 April 2024
- Lab
- xAI
- Kind
- model
- Access
- research preview
Figures
| Measure | Value | Measured by |
|---|---|---|
| RealWorldQA | 68.7% vs GPT-4V 61.4%; benchmark introduced by xAI in the same post | company |
Wikipedia's model table lists Grok-1.5V as unreleased; xAI's post says it 'will be available soon'. Treat as an announced preview. Benchmark was authored by the model's own developer.
Sources
This record was checked against its sources on 6 October 2026. How we check