AI Briefing
KOSign in
Pick

Qwen3.8 27B Records 52 Points on Artificial Analysis Intelligence Index

·2026.08.18 09:32

Key point

Alibaba's Qwen3.8 27B model recorded 52 points on the Artificial Analysis Intelligence Index.

Details

Alibaba's open-weight model, Qwen3.8 27B, achieved 52 points on the Artificial Analysis Intelligence Index evaluation.

This evaluation was conducted by integrating the following 9 refined evaluation items:

  • Coding and Agent Capabilities: Terminal-Bench v2.1 (coding and terminal usage), SciCode (coding), GDPval-AA v2 (agent practical tasks)
  • Reasoning and Scientific Knowledge: GPQA Diamond, CritPt (physics reasoning), Humanity's Last Exam (reasoning and knowledge)
  • Knowledge Reliability and Context: AA-Omniscience (knowledge accuracy and hallucination prevention), AA-LCR (long-context reasoning)

Artificial Analysis provides comprehensive analysis of Speed, Cost, and Openness Index, in addition to the model's intelligence, through these metrics.

qwenalibabamodel-evaluationartificial-analysisllm-benchmark
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

RedditQwen Reaches 167M Monthly Users as Alibaba Reports 250M AI Shopping TrialsRedditMeta's 816B Muse Spark Open scores 39 on Artificial Analysis, trailing 180B Qwen3.8-Flash-NextGeekNewsStarSkirmish Bench v0.2: GPT-6 Astra Leads LLM StarCraft Bot Coding, Human Bots Still DominateHacker NewsCohere Reports Qwen3.8-Flash-Next Attempting Unauthorized Network Uploads in SandboxHacker NewsArtificial Analysis Intelligence Index v5: Claude Opus 5.5 Leads as Open Weights Gap Widens to 12 PointsGeekNewsKimi K3.1 Pro Scores 55 on Artificial Analysis Intelligence Index