Grok Beta: benchmark results
xAI's grok-beta, a provisional 128K-context model offered through its public API beta in 2024. Provider: xAI. Released 2024-10-21. Access: API.
Aggregate Rating 1587 ± 20, rank #487 of 1656 rated models, from 88 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Language - NLP - English (CompassBench 2411) | 63.4 | Score (%) | 100 |
| OpenCompass LLM - Language - English (CompassBench 2411) | 60.7 | Score (%) | 97 |
| OpenCompass LLM - Reasoning - English (CompassBench 2411) | 59.8 | Score (%) | 97 |
| OpenCompass Reasoning - Science and Engineering - English (CompassBench 2411) | 64.5 | Score (%) | 97 |
| SpeechMap Compliance | 93.7 | % Requests Completed | 96.4 |
| OpenCompass Reasoning - Social - English (CompassBench 2411) | 72 | Score (%) | 93.9 |
| ForecastBench Baseline - Zero-Shot - Market | 61.2 | Market Brier Index (difficulty-adjusted, higher is better) | 92.5 |
| TextClass Benchmark | 1741.94 | Meta-Elo (imported) | 91.6 |
| OpenCompass Code - Code QA - English (CompassBench 2411) | 58.2 | Score (%) | 90.9 |
| OpenCompass LLM - Reasoning (CompassBench 2411) | 58.6 | Score (%) | 90.9 |
| OpenCompass Reasoning - Humanities - English (CompassBench 2411) | 64.4 | Score (%) | 90.9 |
| ProLLM - Function Calling | 91.8 | Score (%) | 90.7 |
Interactive version: theaggregate.ai/model?slug=grok-beta · How It Works · Data refreshed daily, snapshot 2026-10-20.