Benchmark
Indigenous Americas Language Identification Full (zero-shot)
Scores are percentages. Higher is better.
- Models measured
- 7
- Scores recorded
- 7
- Sources
- 1
- IQ weight
- 1.50
Score distribution
- 42.3%
- Mean
- 37%
- Median
- 15%
- Std dev
- 30%
- 10th percentile
- 63%
- 90th percentile
- 24%
- Min
This benchmark has a weight of 1.50 for chat models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.0 Flashtable: 1, mode: Full· Indigenous Americas linguistic competence data docs | 65% |
| 2 | GPT 4.1table: 1, mode: Full· Indigenous Americas linguistic competence data docs | 63% |
| 3 | DeepSeek V3table: 1, mode: Full· Indigenous Americas linguistic competence data docs | 45% |
| 4 | Qwen 2.5 14B Instructtable: 1, mode: Full· Indigenous Americas linguistic competence data docs | 37% |
| 5 | Llama 3.1 8B Instructtable: 1, mode: Full· Indigenous Americas linguistic competence data docs | 32% |
| 6 | Qwen2.5 7B Instructtable: 1, mode: Full· Indigenous Americas linguistic competence data docs | 30% |
| 7 | Qwen2.5 3B Instructtable: 1, mode: Full· Indigenous Americas linguistic competence data docs | 24% |
Data credits: Vasselli, Martínez Peguero, Hudi, Sakajo and Watanabe. Measuring Linguistic Competence of LLMs on Indigenous Languages of the Americas. EACL 2026, ©2026 ACL, CC BY 4.0. Numeric/protocol extract from Tables 1–4; random reference rows omitted. (licence)