Grok-3
X.AI·
LLMsflagshipcloud
Overview
Massive scale model trained on the world's largest supercomputer.
Capabilities and innovations
100k H100 trainingSuper-intelligenceMultimodal
Benchmarks
| Benchmark | Score | Source |
|---|---|---|
| GPQA Diamond | 81.2% | unsourced |
| Humanity’s Last Exam (no tools) | 25.6% | unsourced |
| SWE-bench Verified | 51.8% | unsourced |
| MMLU | 86.5% | unsourced |
- SWE-bench Pro: benchmark did not exist at release
- MMLU-Pro: not reported by the vendor
- Terminal-Bench 2.x: benchmark did not exist at release
- MMMU-Pro: benchmark did not exist at release
Links
More from X.AI
Data curated by AI Model Timeline. See the methodology for admission criteria, benchmark eras and source priority.