Grok 4.5vsGLM-5

Grok 4.5
GLM-5
Specifications
Parameters
744B
Benchmarks
Nonsense detection
BullshitBench v2
54%Best
28%
Prompt injection robustness
Gray Swan IPI · k = 1
13.4%
Prompt injection robustness
Gray Swan IPI · k = 10
54.2%
Prompt injection robustness
Gray Swan IPI · k = 15
60.8%
Agentic coding
SWE-Bench Pro
64.7%
Coding
SWE-Bench Verified
77.8%
Multilingual coding
SWE-Bench Multilingual
78%Best
73.3%
Agentic coding
CursorBench v3.2
66.7%
Agentic coding
DeepSWE 1.1
54%
Agentic coding
DeepSWE 1.0
62%
Next.js coding
Next.js Evals
83%
Agentic computer work
Frontier-Bench v0.1
17.8%
Agentic terminal coding
Terminal-Bench 2.1
83.3%
Agentic terminal coding
Terminal-Bench 2.0
56.2%
Web browsing
BrowseComp
75.9%
Multidisciplinary reasoning
Humanity's Last Exam · with tools
50.4%
Science
GPQA Diamond
86%
Agentic legal work
Harvey's Legal Agent Benchmark
12.92%
Medical admin work
MedScribe
86.88%
Community preference
Arena Elo (Text)
1468
Community preference (code)
Arena Elo (Code)
1549Best
1435
Overview
CompanySpaceXAIZ.ai
Release dateJul 8 2026Feb 12 2026
AccessProprietaryOpen Weight

Which is better: Grok 4.5 or GLM-5?

Grok 4.5 leads GLM-5 on 3 of the 3 benchmarks they both report (BullshitBench v2, SWE-Bench Multilingual, Arena Elo (Code)). GLM-5 shipped 146 days before Grok 4.5, so benchmark comparisons should account for the intervening progress.

Grok 4.5 is proprietary, while GLM-5 is open weight.

On BullshitBench v2, Grok 4.5 leads at 54% vs GLM-5 at 28%. On SWE-Bench Multilingual, Grok 4.5 leads at 78% vs GLM-5 at 73.3%. On Arena Elo (Code), Grok 4.5 leads at 1549 vs GLM-5 at 1435.

Frequently asked questions

Grok 4.5 was released by SpaceXAI on Jul 8 2026.

GLM-5 was released by Z.ai on Feb 12 2026.

Grok 4.5 leads on SWE-Bench Multilingual — Grok 4.5 78% vs GLM-5 73.3%.

Grok 4.5 is a proprietary model released by SpaceXAI. GLM-5 is an open weight model released by Z.ai.

Other comparisons