Quality & Security Auto-verified translation
Technology
GLM-5.3
GLM-5.3 — Technology. definition: Z.ai model announced on August 14, 2026, built on the same base model as GLM-5.2 — all the gain comes from post-training. Claims the open-weights state of the art on Terminal-Bench 3.0 (28.3 vs 17.4 for Kimi K3) and Agents' Last Exam (28.5), +50 % on the internal Z.ai Code Bench benchmark, and the state of the art across all models on CyberGym (84.5 %). Three reasoning effort levels (low, high, max, default max); disabling reasoning is no longer supported. Weights announced for two weeks after launch
- Type
- Technology
- definition
- Z.ai model announced on August 14, 2026, built on the same base model as GLM-5.2 — all the gain comes from post-training. Claims the open-weights state of the art on Terminal-Bench 3.0 (28.3 vs 17.4 for Kimi K3) and Agents' Last Exam (28.5), +50 % on the internal Z.ai Code Bench benchmark, and the state of the art across all models on CyberGym (84.5 %). Three reasoning effort levels (low, high, max, default max); disabling reasoning is no longer supported. Weights announced for two weeks after launch
- relations
- 16
- Cited in
- 1 fiches
Adoption measures
-
28.3 on Terminal-Bench 3.0 vs 4.6 for GLM-5.2, 66.9 on DeepSWE v1.1 vs 46.2, and 28.5 on Agents' Last Exam vs 23.8
-
84.5 % on CyberGym vs 77.2 % for GLM-5.2, 54.4 % on ExploitBench vs 24.4 %, and 105 then 130 ExploitGym tasks under normalized 2h and 6h budgets vs 29 and 39
-
34.5 % on Z.ai Code Bench at about 75,000 output tokens per task in Max effort, vs 23.4 % at 96,000 for GLM-5.2, and 31.4 % at about 50,000 tokens in High effort
-
2,436 vulnerabilities identified in 269 open source projects after expert review, triage and deduplication, covering system kernels, operating systems, browser engines, open source infrastructure, web applications and network protocols
Neighborhood
→ is a variant of
GLM TECHNOLOGIE high confidence stable
→ uses
slime TECHNOLOGIE high confidence stable Source ↗
SAO METHODOLOGIE high confidence stable Source ↗
IndexShare TECHNOLOGIE high confidence stable Source ↗
→ observed in
Claude Code TECHNOLOGIE high confidence stable Source ↗
→ outperforms
Claude Opus 4.8 TECHNOLOGIE high confidence stable Source ↗