Icelandic LLM Leaderboard - Average: leaderboard
Mideind's average over seven Icelandic benchmarks built from ten tasks: Wikipedia questions, song lyrics and famous phrases, textile terms and species names, generative quality, error correction, case government, and idiom completion and meaning. Five tasks are scored by exact match, three by an LLM judge, lyrics by overlap and generated text by defect checks. The average is a percentage, higher is better.
Metric: Average Score (%). Source: huggingface.co. Saturation forecast: Around February 2027. 70 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 5.5 | 70.87 | #3 |
| 2 | Gemini 4 Argon | 70.52 | #8 |
| 3 | GPT-6 Astra | 70.25 | #6 |
| 4 | Claude Fable 5.1 | 69.42 | #9 |
| 5 | Claude Opus 5 | 68.63 | #12 |
| 6 | GPT-6.1 Sol | 68.17 | #7 |
| 7 | Gemini 3.1 Pro (Preview) | 66.54 | #58 |
| 8 | GPT-6 Sol | 65.63 | #14 |
| 9 | GPT-5.6 Sol | 65.29 | #19 |
| 10 | Gemini 3.7 Flash | 64.97 | #23 |
| 11 | Claude Opus 4.8 | 63.88 | #37 |
| 12 | Gemini 3.8 Flash | 63.48 | #21 |
| 13 | Gemini 3.5 Flash | 62.22 | #60 |
| 14 | Claude Sonnet 5.5 | 62.12 | #11 |
| 15 | GPT-6 Luna | 57.14 | #56 |
No result here: #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=icelandic-llm-leaderboard-average · How It Works · Data refreshed daily, snapshot 2026-10-20.