Z.aiEvidence ready

GLM 5.2

Z.ai / GLM 5.2 · zai-org/GLM-5.2

Chinese open-weight long-horizon agentic engineering model with 1M-token context.

Objective score

52.3

Confidence

89%

Price evidence

Open weights; hosted pricing varies

Support this model Compare trend Official source

Other primary tiers in this family

GLM 4 32B 0414 GLM 4 9B 0414 GLM 4.5 GLM 4.5 Air GLM 4.6 GLM 4.7 GLM 5 GLM 5.1GLM 5.2GLM Z1 32B 0414 GLM Z1 9B 0414 GLM Z1 Rumination 32B 0414

Scenario evidence

Six task-specific rankings

Feature development#5Obj. 52.30 · 0%Debug and repair#9Obj. 53.30 · 0%Refactoring#13Obj. 49.30 · 0%Code review#11Obj. 52.90 · 0%Large-repo agent work#4Obj. 52.60 · 0%Frontend / UI#19Obj. 48.10 · 0%

Evidence state

Signal coverage

Benchmarks: 4
Rank points: 7
Usage: 4
Agent: 1
Sources: 4

Objective evidence

Benchmarks and public signals

Swebench: 62.1
Livecodebench: —
Aider: —
Artificial Analysis: 68.8
Openrouter Usage: 10.1
Kilo Usage: 56

Persistent history

Recent ranking points

Aug 1#10 · 49

Aug 1#10 · 49

Jul 31#10 · 51

Jul 31#10 · 51

Jul 31#9 · 51

Jul 31#9 · 51

Usage signals

Kilo56%

Overall51.7%

OpenRouter coding12.1%

OpenRouter10.1%

Agent and tool signals

Claude Code · 43.2

Claude Code - GLM-5.2

Source trail

Official GLM-5 repo describes GLM-5.2 coding gains and 1M-token context.Artificial Analysis Coding Agents · ok WebDev Arena · ok Kilo leaderboard · ok GLM-5 repository · ok