Kagi LLM Benchmark: leaderboard
Rotating set of unpublished reasoning, coding and instruction-following tasks run by Kagi; accuracy on ~116 private tasks designed to resist benchmark contamination.
Metric: Accuracy (%). Source: help.kagi.com. Saturation forecast: Around January 2027. 145 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 88.8 | #31 |
| 2 | Claude Fable 5 | 88.8 | #16 |
| 3 | Claude Opus 4.8 (Thinking) | 88.8 | #37 (Claude Opus 4.8) |
| 4 | Claude Opus 4.6 (Thinking) | 83.6 | #63 (Claude Opus 4.6) |
| 5 | Grok 4.5 | 83.5 | #48 |
| 6 | Claude Opus 4.7 (Thinking) | 80.7 | #49 (Claude Opus 4.7) |
| 7 | Claude Opus 4.5 (Thinking) | 80.2 | #83 (Claude Opus 4.5) |
| 8 | Gemini 3 Pro | 80.1 | #80 |
| 9 | Kimi K2.5 (Thinking) | 78.5 | #144 (Kimi K2.5) |
| 10 | GPT-5 Pro | 76.8 | #81 |
| 11 | GLM-5 (Thinking) | 75 | #141 (GLM-5) |
| 12 | Grok 4.20 (Reasoning) | 75 | #175 (Grok 4.20) |
| 13 | Claude Opus 4 (Thinking) | 74.3 | #163 (Claude Opus 4) |
| 14 | Grok 4 | 73.6 | #181 |
| 15 | GPT-5.2 | 73.3 | #110 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kagi-llm-benchmark · How It Works · Data refreshed daily, snapshot 2026-10-20.