The LLM benchmark aggregator.

GLM-4.5

Z.ai · 358B · open weights

Strongest in Mathematics (#47 of 152, on 1 of its 2 boards), weakest in Harm refusal (#251 of 292). Above par in 11 of 19 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4.8 and GPT-5 and ahead of DeepSeek V4 Flash and DeepSeek V3.2.

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Core abilities51.8−13.5#76/1991/4

General intelligence52.5−16.9#64/1781/3

Professional51.4−11#82/1941/2

Medical53.4−13.1#62/1491/1

Finance52.9−9.5#75/1621/1

Legal48−17.9#106/1731/1

Reasoning51.7−13.7#85/1951/5

Mathematics55.4−8.4#47/1521/2

Science47.3−16.2#85/1221/1

Knowledge50−12.5#113/1971/3

Academic knowledge50−13.9#79/1231/1

Human preference57.7−11.1#124/3541/1

Safety48.2−13.2#230/3291/3

Fairness55.4−14.6#64/2921/2

Toxicity avoidance52.9−6.1#152/2681/1

Secure code48.2−18#174/2701/1

Jailbreak resistance44.1−20.9#203/2681/1

Harm refusal44.4−17.1#251/2921/2

§ 2 · Head to head

What it beats, and what beats it

The same models turn up scope after scope. Counted once: where both were placed, who finished higher. Bars run right for GLM-4.5, left for the other.

Magistral Small 1.2won 14 · lost 1

Jamba 1.6 Miniwon 13 · lost 2

Jamba 1.6 Largewon 13 · lost 2

Command R+won 13 · lost 2

GPT-5.4 nanowon 12 · lost 2

Mistral Medium 3.5won 11 · lost 2

DeepSeek V3won 16 · lost 3

Jamba 1.5 Largewon 10 · lost 2

Gemini 2.5 Flash-Litewon 13 · lost 3

GPT OSS 20Bwon 15 · lost 4

Llama 4 Scout Instructwon 15 · lost 4

Command Awon 15 · lost 4

Gemini 2.0 Flashwon 15 · lost 4

Mistral Large 3won 11 · lost 3

Gemini 1.5 Prowon 13 · lost 4

Jamba 1.5 Miniwon 9 · lost 3

Gemini 1.5 Flashwon 12 · lost 4

GPT-4.1 nanowon 14 · lost 5

Llama 4 Maverick Instructwon 14 · lost 5

GPT-4.1 miniwon 14 · lost 5

GPT-4o miniwon 13 · lost 5

MiniMax M2.5won 10 · lost 4

Inkling Smallwon 9 · lost 4

Llama 3.3 70B Instructwon 11 · lost 5

Claude 3.5 Haikuwon 11 · lost 5

Kimi K2won 13 · lost 6

Llama 3.1 405B Instructwon 8 · lost 4

MiniMax M2.1won 8 · lost 4

O3 Miniwon 11 · lost 6

DeepSeek V3.1won 9 · lost 5

GPT OSS 120Bwon 12 · lost 7

NVIDIA Nemotron 3 Ultra 550B A55Bwon 12 · lost 7

Gemma 4 31Bwon 9 · lost 6

GPT-4owon 10 · lost 7

GPT-5 nanowon 11 · lost 8

Grok 3won 11 · lost 8

Claude 3.5 Sonnetwon 9 · lost 7

DeepSeek R1won 10 · lost 8

Grok 4.3won 10 · lost 8

Qwen3.8 27Bwon 7 · lost 6

GPT-5.4 miniwon 7 · lost 6

Gemini 3.5 Flash-Litewon 7 · lost 6

MiMo V2.6 Flashwon 8 · lost 7

Claude Haiku 4.5won 9 · lost 8

O4 Miniwon 10 · lost 9

Gemini 2.5 Flashwon 10 · lost 9

DeepSeek V3.2won 10 · lost 9

DeepSeek V4 Flashwon 10 · lost 9

GPT-4 Turbowon 6 · lost 6

Grok 4 Fastwon 7 · lost 7

GPT-4.1won 9 · lost 10

GLM-4.6won 9 · lost 10

GPT-6 Lunawon 8 · lost 9

MiMo V2.5 Prowon 8 · lost 9

Kimi K2.6won 8 · lost 10

Ling 3.0 Flashwon 6 · lost 8

Gemini 3.1 Flash-Litewon 6 · lost 8

Grok 4won 8 · lost 11

Qwen3 235B A22B Instructwon 8 · lost 11

Qwen3.6 Pluswon 5 · lost 7

Grok 4.1 Fastwon 7 · lost 10

Mistral Large 4won 6 · lost 10

Claude Sonnet 4won 7 · lost 12

MiniMax M3won 7 · lost 12

Inklingwon 7 · lost 12

GPT-5.6 Lunawon 7 · lost 12

Claude 3.7 Sonnetwon 6 · lost 11

MiniMax M2.7won 6 · lost 11

GLM-5.1won 6 · lost 11

Qwen3.5 Flashwon 4 · lost 8

DeepSeek V4 Prowon 6 · lost 13

GLM-5.2won 6 · lost 13

MiMo V2.6 Prowon 5 · lost 11

Grok 4.20won 4 · lost 9

Kimi K2 Thinkingwon 5 · lost 12

O1won 5 · lost 12

Qwen3.7 Maxwon 5 · lost 13

Gemini 3.5 Flashwon 5 · lost 13

Gemini 3.6 Flashwon 5 · lost 13

GPT-5 miniwon 5 · lost 14

Kimi K2.5won 5 · lost 14

GLM-5.3 Flashwon 5 · lost 14

Qwen3.8 Maxwon 5 · lost 14

GLM-4.7won 3 · lost 9

Claude Opus 4.1won 3 · lost 9

Grok 3 Miniwon 4 · lost 12

Qwen3 Maxwon 3 · lost 10

Claude Opus 4won 4 · lost 15

O3won 4 · lost 15

GPT-5.6 Terrawon 4 · lost 15

GPT-6 Solwon 3 · lost 13

Claude Sonnet 4.6won 3 · lost 13

Claude Opus 4.6won 3 · lost 14

DeepSeek V4.1 Flashwon 2 · lost 10

Muse Spark 1.1won 3 · lost 15

Gemini 2.5 Prowon 3 · lost 16

Kimi K3won 2 · lost 11

Grok 4.7won 2 · lost 14

Claude Sonnet 5won 2 · lost 16

Grok 4.6won 2 · lost 17

Grok 4.5won 2 · lost 17

GPT-5.4won 2 · lost 17

GPT-5.6 Solwon 2 · lost 17

Gemini 3.8 Flashwon 2 · lost 17

Gemini 3 Flashwon 1 · lost 11

Gemini 3.7 Flashwon 1 · lost 11

GLM-5.3won 1 · lost 12

Gemini 3.1 Prowon 1 · lost 12

GLM-5won 1 · lost 13

GPT-5.1won 1 · lost 16

Claude Sonnet 4.5won 1 · lost 16

GPT-5.2won 1 · lost 18

GPT-5.5won 1 · lost 18

Claude Opus 4.7won 1 · lost 18

GPT-6 Astrawon 0 · lost 12

Claude Fable 5.1won 0 · lost 13

Claude Fable 5won 0 · lost 13

Claude Opus 4.5won 0 · lost 14

Gemini 3 Prowon 0 · lost 16

Claude Opus 5.5won 0 · lost 17

Claude Opus 5won 0 · lost 18

GPT-5won 0 · lost 19

Claude Opus 4.8won 0 · lost 19

§ 3 · Sources

Where the numbers come from

4 publications, 15 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1411

Kagi LLM Benchmark ↗1 measureread 2026-11-10

Kagi LLM Benchmark 57.9%

Vals.ai ↗7 measuresread 2026-11-10

Vals · LegalBench 75.63%Vals · CorpFin 60.96%Vals · TaxEval 72.4%Vals · MedQA 89.97%Vals · GPQA Diamond 72.22%Vals · MMLU Pro 81.22%Vals · AIME 86.67%

Enkrypt AI Safety Leaderboard ↗6 measuresread 2026-11-10

Enkrypt · Jailbreak risk 17.8%Enkrypt · Harmful content risk 47.8%Enkrypt · CBRN risk 22.3%Enkrypt · Toxicity risk 4.2%Enkrypt · Bias risk 69.5%Enkrypt · Insecure code risk 36.9%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=glm-4-5)](https://publicai.io/model-index/m/glm-4-5)