Linguini - IOL-AI Harness: leaderboard
The Linguini olympiad-linguistics set re-run with the IOL-AI Challenge harness on a 2026 model panel: geometric mean of chrF and exact match after answer parsing. Higher is better. Source: arxiv.org/abs/2608.18011, Table 12.
Metric: Geometric Mean of chrF and Exact Match (%; after custom answer parsing). Source: arxiv.org. Saturation forecast: Around February 2027. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.6 Sol | 51.5 | #29 |
| 2 | Gemini 3.6 Flash | 50.8 | #74 |
| 3 | Claude Opus 4.8 | 43.7 | #57 |
| 4 | Qwen 3.7 Max | 35.6 | #143 |
| 5 | DeepSeek V4 Pro | 32.9 | #188 |
| 6 | GLM-5.2 | 30.7 | #138 |
| 7 | Kimi K2.6 | 28.8 | #176 |
| 8 | Inkling | 28.6 | #271 |
| 9 | Qwen 3.8 27B | 25.2 | #206 |
| 10 | Muse Glimmer 30B (High) | 22.9 | #381 (Muse Glimmer 30B) |
| 11 | Nemotron 3 Ultra | 19.4 | #427 |
| 12 | Gemma 4 31B | 17.8 | #392 |
| 13 | Qwen 3.6 27B (Thinking) | 15.9 | #345 (Qwen 3.6 27B) |
| 14 | Qwen 3.6 27B (Non-reasoning) | 15.8 | #345 (Qwen 3.6 27B) |
| 15 | MiniMax-M3 | 11.2 | #270 |
No result here: #4 Claude Opus 5.5, #5 GPT-6 Astra, #9 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=linguini-iol-ai-harness · How It Works · Data refreshed daily, snapshot 2026-11-08.