OpenAI answer share
share of recommendation answers52.8%+5.6 pts
LiveBench data, Rank.ai analysis
Best AI model score on LiveBench
The highest overall score on LiveBench, an objective benchmark with contamination-resistant questions, held by the current leading model configuration.
LiveBench overall score82.4held by GPT 5.6 Sol Max
Behind the number
The full field
- GPT 5.6 Sol Max82.4
- Claude Fable 5 Max Effort80.8
- Claude Opus 5 xHigh Effort80.3
- GPT 5.5 xHigh79.9
- GPT 5.6 Terra Max79.8
- GPT 5.6 Sol xHigh79.7
- Claude Fable 5 xHigh Effort79.5
- Claude Opus 5 Max Effort79.2
- Claude Opus 5 High Effort79.0
- Claude Opus 4 8 xHigh Effort78.9
- GPT 5.5 High78.7
- Kimi K378.5
The board in rank order. The headline number is the top-ranked row, highlighted. Showing the top 12 of 45 ranked entries.
| Rank | Entry | LiveBench overall score |
|---|---|---|
| 1 | GPT 5.6 Sol Max | 82.4 |
| 2 | Claude Fable 5 Max Effort | 80.8 |
| 3 | Claude Opus 5 xHigh Effort | 80.3 |
| 4 | GPT 5.5 xHigh | 79.9 |
| 5 | GPT 5.6 Terra Max | 79.8 |
| 6 | GPT 5.6 Sol xHigh | 79.7 |
| 7 | Claude Fable 5 xHigh Effort | 79.5 |
| 8 | Claude Opus 5 Max Effort | 79.2 |
| 9 | Claude Opus 5 High Effort | 79.0 |
| 10 | Claude Opus 4 8 xHigh Effort | 78.9 |
| 11 | GPT 5.5 High | 78.7 |
| 12 | Kimi K3 | 78.5 |
82.4 on every publication so far. A line appears here as soon as the number moves.
LiveBench data, Rank.ai analysis
About this metric
LiveBench is an objective benchmark that scores exact model configurations across seven categories, including reasoning, coding, mathematics, and data analysis. Questions rotate to resist contamination, and grading is automatic rather than judged by other models. This chart tracks the single best overall score on the board, whichever configuration holds it.
The score is the unweighted mean of the seven LiveBench category averages for each official release, republished here per exact configuration rather than per marketing name. A model listed at high effort or maximum thinking budget is scored as that configuration only. The leader shown beside the number is read from the latest snapshot.
Vendor announcements lead with the evaluations they choose; a rotating, objective benchmark scores every model the same way. Watching the top score move, and which model holds it, is a simple way to follow capability between announcements.
Charts
Related charts
See how often AI names you.
Enter your website. In about two minutes, Rank.ai asks ChatGPT, Claude and Gemini 12 questions your customers ask and grades how often they name you.
- Your grade out of 100How often AI names you, cites your site, and how high it ranks you.
- Who gets namedEvery competitor in the answers, most named first.
- The pages AI readsThe sources behind each answer.
- Three fixesWhat to fix first, with a brief for the first article.