Qwen3.8-27BvsGLM-5.2

Qwen3.8-27B
GLM-5.2
Specifications
Parameters
27B
744B
Context window
262k
1M
Benchmarks
Nonsense detection
BullshitBench v2
31%
Agentic coding
SWE-Bench Pro
61.7%
62.1%Best
Agentic coding
CursorBench v3.2
55%
Agentic coding
CursorBench v3.1
54.6%
Agentic coding
DeepSWE 1.1
42.2%
44%Best
Next.js coding
Next.js Evals
88%
Repo-level code generation
NL2Repo-Bench
42.3%
Software engineering
QwenSWEBench
79%
Competitive coding
LiveCodeBench
90.3%
Agentic computer work
Frontier-Bench v0.1
5.1%
Agentic terminal coding
Terminal-Bench 2.1
73%
81%Best
Professional tool use
JobBench
33.4%
Long-horizon office work
CoWorkBench
70.7%
Multidisciplinary reasoning
Humanity's Last Exam · no tools
30.8%
40.5%Best
Multidisciplinary reasoning
Humanity's Last Exam · with tools
54.7%
Science
GPQA Diamond
89.2%
91.2%Best
Instruction following
IFBench
79.5%
Agentic computer use
Agent's Last Exam · pass@1
20.4%
Agentic computer use
Agent's Last Exam · score
42.9%
Knowledge work
GDPval-AA v2
1514
Community preference (code)
Arena Elo (Code)
1587
Overview
CompanyQwenZ.ai
Release dateAug 14 2026Jun 16 2026
AccessOpen WeightOpen Weight

Which is better: Qwen3.8-27B or GLM-5.2?

GLM-5.2 leads Qwen3.8-27B on 5 of the 5 benchmarks they both report. GLM-5.2 shipped 59 days before Qwen3.8-27B, so benchmark comparisons should account for the intervening progress.

Qwen3.8-27B has 27B parameters, while GLM-5.2 has 744B. Context windows are 262k (Qwen3.8-27B) vs 1M (GLM-5.2).

On SWE-Bench Pro, GLM-5.2 leads at 62.1% vs Qwen3.8-27B at 61.7%. On DeepSWE 1.1, GLM-5.2 leads at 44% vs Qwen3.8-27B at 42.2%. On Terminal-Bench 2.1, GLM-5.2 leads at 81% vs Qwen3.8-27B at 73%. On Humanity's Last Exam · no tools, GLM-5.2 leads at 40.5% vs Qwen3.8-27B at 30.8%. On GPQA Diamond, GLM-5.2 leads at 91.2% vs Qwen3.8-27B at 89.2%.

Frequently asked questions

Qwen3.8-27B was released by Qwen on Aug 14 2026.

GLM-5.2 was released by Z.ai on Jun 16 2026.

GLM-5.2 leads on SWE-Bench Pro — Qwen3.8-27B 61.7% vs GLM-5.2 62.1%.

GLM-5.2 leads on Humanity's Last Exam · no tools — Qwen3.8-27B 30.8% vs GLM-5.2 40.5%.

Qwen3.8-27B has a 262k context window; GLM-5.2 has 1M.

Other comparisons