Yi 1.5 34B Chat — benchmark results

01.AI's 34B chat model from the Yi-1.5 series, continue-pretrained on 500B extra tokens for stronger coding, math, reasoning and instruction following (May 2024). Provider: 01.AI. Released 2024-05-13. Access: Open.

Unified ELO 1467 ± 28, rank #923 of 1776 rated models, from 59 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Open CoT - LSAT Logical Reasoning24.71CoT Gain (%)100
Open Chinese LLM - GSM8K68.31Accuracy (%)92.9
Open LLM Leaderboard - GPQA15.32Score91.6
Open Chinese LLM - ARC Challenge61.6Accuracy (%)90.9
Open Chinese LLM Leaderboard68.17Average Score (%)90.2
Open CoT Leaderboard13.81Average CoT Gain (%)90.1
Open CoT - LogiQA 214.19CoT Gain (%)88.2
Open CoT - LSAT Reading Comprehension20.45CoT Gain (%)87
Open Chinese LLM - WinoGrande68.27Accuracy (%)86.5
Open Chinese LLM - C-Eval Semantic87.44Accuracy (%)85.3
Open LLM Leaderboard - BBH44.26Score84.5
Open LLM Leaderboard - MMLU-Pro39.12Score84.1

Interactive version: theaggregate.ai/model?slug=yi-1-5-34b-chat · How the rankings work · Data refreshed daily, snapshot 2026-07-22.