GLM-5.3 benchmarks and performance, in charts
Every chart we tracked that shows GLM-5.3, from the lab itself and from the trackers people trust, with how far each one travelled.
GLM-5.3 on its own benchmark table
Drawn from the numbers huggingface.co published as text (Benchmark, read Oct 12, 2026). Each panel is one test; the highlighted bar is GLM-5.3. Source: huggingface.co.
See the numbers
| Terminal Bench 2.1 | GPT-5.6 Sol 88.8, Kimi K3 88.3, GLM-5.3 88.2, Fable 5 (w/ fallback) 88, DeepSeek-V4 Pro-0813 87.9, Qwen3.8-Max 86.6, Opus 4.8 85, GLM-5.2 81 |
|---|---|
| Terminal Bench 3.0 | GPT-5.6 Sol 34.6, Fable 5 (w/ fallback) 33.7, GLM-5.3 28.3, Opus 4.8 21.1, Kimi K3 17.4, GLM-5.2 4.6 |
| DeepSWE (v1.1) | GPT-5.6 Sol 72.7, Fable 5 (w/ fallback) 69.7, Kimi K3 67.5, GLM-5.3 66.9, DeepSeek-V4 Pro-0813 62.7, Opus 4.8 58, Qwen3.8-Max 56.6, GLM-5.2 46.2 |
| NL2Repo | Opus 4.8 69.7, DeepSeek-V4 Pro-0813 61.1, GLM-5.3 58, Kimi K3 58, Qwen3.8-Max 55.9, GLM-5.2 48.9 |
| ProgramBench (Almost Solved) | Fable 5 (w/ fallback) 33, GPT-5.6 Sol 23, GLM-5.3 19, Kimi K3 17.5, Opus 4.8 15.5, Qwen3.8-Max 10.5, GLM-5.2 9.5 |
| FrontierSWE | Fable 5 (w/ fallback) 88.2, GLM-5.3 78.1, GLM-5.2 67.5, Opus 4.8 66.5 |
Official charts from Z.ai
As published on X, Hugging Face, GitHub and their blog. Each card opens our own page for the chart.
From benchmark trackers and others
More charts that feature GLM-5.3
Benchmark trackers, other labs and researchers. Each card opens our own page for the chart.
%201_Z2dNL7P.webp?dpl=6accf0d261fecbe6ba702f4d)
Free with your email
See all 31 GLM-5.3 charts
Every official launch chart and independent benchmark we found for GLM-5.3, with who posted each one. Plus Monday's email: the week's charts that beat their usual likes.
Unlocks all 35 giants. One email a week, leave any time. What you get · Log in
More charts that feature GLM-5.3 more
%201_1fqgOj.webp?dpl=6accf0d261fecbe6ba702f4d)
Make it yours
Your numbers, in a look-alike of the original.
Starts from the real numbers of "LLM Performance Bench 3.0" by Z.ai. Replace every number with yours before you post. Make one like this, step by step
Look-alike score 9.2 of 10. Free pictures carry a small "Made with ViralChart" line. That line is how other people find us.
Questions people ask
How does GLM-5.3 perform?
We hold 4 charts that show GLM-5.3, posted between Sep 2026 and Oct 2026. The most seen is "AI model cost performance" by Cursor with 255,335 views.
Who publishes GLM-5.3 benchmark charts?
Cursor, SemiAnalysis, Artificial Analysis, Social Capital, Z.ai, @arena, @MistralAI, mistral.ai, openrouter