Kagi LLM Benchmark: leaderboard
Metric: Accuracy (%). Source: help.kagi.com. 145 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 88.8 |
| 2 | Claude Fable 5 | 88.8 |
| 3 | Claude Opus 4.6 (Thinking) | 83.6 |
| 4 | Grok 4.5 | 83.5 |
| 5 | Claude Opus 4.7 (Thinking) | 80.7 |
| 6 | Claude Opus 4.5 (Thinking) | 80.2 |
| 7 | Gemini 3 Pro | 80.1 |
| 8 | Kimi K2.5 (Thinking) | 78.5 |
| 9 | GPT-5 Pro | 76.8 |
| 10 | GLM-5 (Thinking) | 75 |
| 11 | Grok 4.20 (Reasoning) | 75 |
| 12 | Claude Opus 4 (Thinking) | 74.3 |
| 13 | Grok 4 | 73.6 |
| 14 | GPT-5.2 | 73.3 |
| 15 | Claude Opus 4.7 | 73.3 |
Interactive version: theaggregate.ai/benchmark?slug=kagi-llm-benchmark · How It Works · Data refreshed daily, snapshot 2026-09-05.