AnFeng_v3.1-Avocet: benchmark results
NLPark's sparsely documented 34B conversational model on the Yi-34B architecture, tagged not-for-all-audiences. Provider: NLPark. Released 2024-08-03. Access: Open.
Aggregate Rating 1422 ± 22, rank #1138 of 1654 rated models, from 20 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open Korean LLM - Ko-GPQA-Diamond | 27.78 | Normalized accuracy (%) | 95 |
| Open Portuguese LLM - FaQuAD NLI | 81.63 | Macro F1 (%) | 92.6 |
| Open LLM Leaderboard - MMLU-Pro | 44.38 | Score | 82.3 |
| Open LLM Leaderboard - MuSR | 44.76 | Score | 82.2 |
| Open Portuguese LLM - BLUEX | 61.75 | Accuracy (%) | 81 |
| Open LLM Leaderboard - BBH | 58.29 | Score | 79.4 |
| Open Korean LLM - KorNAT-Harmless | 64.96 | Normalized accuracy (%) | 78.2 |
| Open LLM Leaderboard - GPQA | 32.47 | Score | 78 |
| Open Portuguese LLM - OAB Exams | 50.11 | Accuracy (%) | 75.9 |
| Open Portuguese LLM - ENEM | 69.14 | Accuracy (%) | 72.3 |
| Open Korean LLM - Ko-EQ-Bench | 25.25 | EQ-Bench score | 71.1 |
| Open Korean LLM - Ko-IFEval | 34.54 | Strict accuracy, prompt/instruction mean (%) | 71.1 |
Interactive version: theaggregate.ai/model?slug=anfeng-v3-1-avocet · How It Works · Data refreshed daily, snapshot 2026-10-24.