Yi-1.5-6B: benchmark results

01.AI's Apache-2.0 6B base model (May 2024): Yi-6B continued-pretrained on 500B additional tokens for stronger coding, math, and reasoning. Provider: 01.AI. Released 2024-05-11. Access: Open.

Unified ELO 1398 ± 1, rank #1141 of 1392 rated models, from 57 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open LLM Leaderboard - MuSR13.31Score73.4
EuroEval Portuguese NLU - MultiWikiQA PT72.84Reading comprehension Score (%)72.6
Open Chinese LLM - WinoGrande65.04Accuracy (%)70.8
Open LLM Leaderboard - GPQA8.5Score70.7
MERA - ruCodeEval21.77pass@1 (%)66
Open Chinese LLM - CMMLU57.54Accuracy (%)64.7
Open Chinese LLM - C-Eval Semantic72.2Accuracy (%)58.8
EuroEval Dutch NLU - DBRD87.76Sentiment classification Score (%)54.3
MERA - ruHumanEval13.54pass@1 (%)53.4
EuroEval Portuguese NLU - SST-2 PT77.09Sentiment classification Score (%)51.1
Open Chinese LLM - HellaSwag60.24Accuracy (%)50.4
MERA - ruMultiAr30.08EM (%)46.9

Interactive version: theaggregate.ai/model?slug=yi-1-5-6b · How It Works · Data refreshed daily, snapshot 2026-09-05.