LLM Stats (Toolathlon): leaderboard
LLM Stats vendor-reported Toolathlon rows without an explicit Verified declaration. Exact versions remain unspecified unless established by a primary source; no Original-version default is inferred. Explicit Verified rows are preserved on the separate Verified protocol board. Higher score is better.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around July 2027. 30 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.8 | 59.9 | #37 |
| 2 | GPT-5.6 Sol | 58 | #19 |
| 3 | Gemini 3.5 Flash | 56.5 | #60 |
| 4 | GPT-5.5 | 55.6 | #31 |
| 5 | GPT-5.4 | 54.6 | #79 |
| 6 | Claude Sonnet 5 | 54.3 | #57 |
| 7 | GPT-5.6 Luna | 53.4 | #68 |
| 8 | GPT-5.6 Terra | 53.1 | #41 |
| 9 | DeepSeek V4 Pro (Max) | 51.8 | #102 (DeepSeek V4 Pro) |
| 10 | Seed 2.1 Pro | 50.6 | #67 |
| 11 | Kimi K2.6 | 50 | #104 |
| 12 | Gemini 3 Flash | 49.4 | #99 |
| 13 | Seed 2.1 Turbo | 49.1 | #86 |
| 14 | Hy3 | 48.5 | #93 |
| 15 | GLM-5.2 | 48.2 | #73 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-toolathlon · How It Works · Data refreshed daily, snapshot 2026-10-20.