Qwen 3.5 27B (Reasoning) — benchmark results

Provider: Alibaba. Released 2026-02-24. Access: Open.

Unified ELO 1606 ± 21, rank #518 of 1806 rated models, from 52 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA IFBench75.58Accuracy (%)93.4
AA TAU-2 Bench93.86Accuracy (%)91.1
AA-LCR67.3Score (self-reported)88
AA Long Context Reasoning72.33Accuracy (%)85.7
AA GPQA Diamond85.76Accuracy (%)84.1
AA Global-MMLU-Lite - Japanese89.33Accuracy (%)80.9
AA Humanity's Last Exam23.91Accuracy (%)80.8
Artificial Analysis Intelligence Index34.6Intelligence Index80.3
AA Global-MMLU-Lite - Burmese83.83Accuracy (%)79.1
AA Global-MMLU-Lite - Arabic87.08Accuracy (%)78
AA Global-MMLU-Lite - Indonesian89.5Accuracy (%)77.8
AA Global-MMLU-Lite - Yoruba71.5Accuracy (%)77.5

Interactive version: theaggregate.ai/model?slug=qwen-3-5-27b-reasoning · How It Works · Data refreshed daily, snapshot 2026-08-07.