The LLM benchmark aggregator.

GPT-5.4 mini

OpenAI

Strongest in Factual grounding (#18 of 104), weakest in Core abilities (#185 of 199). Above par in 8 of 21 scopes. Among the models it meets almost everywhere, it finishes behind Claude Opus 4.8 and GPT-5.6 Terra and ahead of Claude Sonnet 4 and GPT-5 mini.

§ 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.

Human preference61.2−7.6#71/3541/1

Safety53.5−7.9#83/3291/3

Factual grounding61.3−9.4#18/1041/1

Knowledge51.7−10.8#90/1971/3

Academic knowledge54.2−9.7#61/1231/1

Reasoning48−17.4#113/1953/5

Science54.9−8.6#59/1221/1

Reasoning45.7−23.3#85/1471/3

Mathematics46−17.8#104/1522/2

Coding44.6−21.6#129/1792/6

Agentic coding44.4−22.5#125/1732/5

Professional47.3−15.1#133/1941/2

Finance51.6−10.8#86/1621/1

Legal38.4−27.5#156/1731/1

Agents49.6−16.8#134/2982/6

Knowledge work49.3−21.4#99/2082/2

Core abilities40.5−24.8#185/1992/4

Language40.7−25.9#55/621/1

Instruction following35.5−38.3#55/621/1

General intelligence42.7−26.7#135/1781/3

§ 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 GPT-5.4 mini, left for the other.

Mistral Medium 3.1won 13 · lost 0

NVIDIA Nemotron 3.5 Lightning 30B A3Bwon 15 · lost 0

GPT-4.1 nanowon 17 · lost 1

GPT-4o miniwon 16 · lost 1

Mistral Medium 3.5won 15 · lost 1

Llama 4 Scout Instructwon 15 · lost 2

Llama 3.3 70B Instructwon 13 · lost 2

Llama 4 Maverick Instructwon 16 · lost 3

Mistral Large 3won 15 · lost 4

GPT-4owon 13 · lost 4

Command Awon 13 · lost 4

Gemini 2.0 Flashwon 12 · lost 4

Gemma 4 31Bwon 12 · lost 4

GPT OSS 20Bwon 11 · lost 4

Claude 3.5 Haikuwon 11 · lost 4

Claude Haiku 4.5won 12 · lost 5

GPT-5 nanowon 11 · lost 5

MiniMax M2.7won 11 · lost 5

Gemini 3.1 Flash-Litewon 12 · lost 6

MiMo V2.5won 9 · lost 5

GPT OSS 120Bwon 12 · lost 7

DeepSeek V3won 12 · lost 7

O3 Miniwon 10 · lost 6

GPT-5.4 nanowon 13 · lost 8

Claude 3.5 Sonnetwon 8 · lost 5

GPT-4.1 miniwon 11 · lost 7

DeepSeek R1won 11 · lost 7

GLM-4.6won 11 · lost 7

Gemini 2.5 Flash-Litewon 9 · lost 6

Claude 3.7 Sonnetwon 9 · lost 6

Grok 4 Fastwon 10 · lost 7

Grok 4.1 Fastwon 9 · lost 7

Kimi K2won 10 · lost 8

MiniMax M2.5won 9 · lost 8

Grok 3won 9 · lost 8

GLM-4.7won 9 · lost 8

GPT-5 miniwon 10 · lost 9

Claude Sonnet 4won 10 · lost 9

Qwen3.5 Flashwon 7 · lost 7

O4 Miniwon 9 · lost 9

Gemini 2.5 Flashwon 9 · lost 9

GPT-4.1won 9 · lost 9

Qwen3 235B A22B Instructwon 9 · lost 9

Grok 4.3won 9 · lost 9

Gemini 2.5 Prowon 9 · lost 10

Gemini 3.5 Flash-Litewon 9 · lost 10

MiniMax M2.1won 7 · lost 8

GLM-4.5won 6 · lost 7

NVIDIA Nemotron 3 Ultra 550B A55Bwon 9 · lost 11

Grok 3 Miniwon 7 · lost 9

O1won 6 · lost 8

DeepSeek V3.2won 8 · lost 11

Claude Opus 4won 7 · lost 10

MiniMax M3won 8 · lost 12

Kimi K2 Thinkingwon 5 · lost 9

Mistral Large 4won 6 · lost 11

Inklingwon 7 · lost 13

MiMo V2.5 Prowon 5 · lost 10

GPT-6 Lunawon 6 · lost 12

Qwen3.8 27Bwon 6 · lost 13

Claude Opus 4.1won 4 · lost 9

Qwen3.6 Pluswon 5 · lost 12

Qwen3 Maxwon 4 · lost 10

Grok 4.20won 4 · lost 11

Gemini 3 Flashwon 4 · lost 11

Kimi K2.5won 5 · lost 14

GLM-5.3 Flashwon 5 · lost 15

Inkling Smallwon 4 · lost 13

Grok 4won 4 · lost 13

Claude Haiku 5.5won 4 · lost 13

Qwen3.7 Maxwon 4 · lost 15

MiMo V2.6 Prowon 3 · lost 12

Claude Opus 4.5won 4 · lost 16

Gemini 3.1 Prowon 4 · lost 16

GPT-6 Solwon 3 · lost 15

GPT-5won 3 · lost 16

Claude Sonnet 4.5won 3 · lost 16

Kimi K2.6won 3 · lost 16

Gemini 3.5 Flashwon 3 · lost 16

Claude Opus 4.6won 3 · lost 16

GPT-5.2won 3 · lost 17

DeepSeek V4 Flashwon 3 · lost 17

Step 5won 2 · lost 12

Muse Sparkwon 2 · lost 12

Gemini 4 Argonwon 2 · lost 14

O3won 2 · lost 15

Grok 4.7won 2 · lost 15

GLM-5won 2 · lost 15

Gemini 3 Prowon 2 · lost 16

GPT-6 Astrawon 2 · lost 17

Claude Sonnet 4.6won 2 · lost 18

DeepSeek V4 Prowon 2 · lost 19

GPT-5.6 Solwon 2 · lost 19

MiMo V2.6 Flashwon 1 · lost 13

GLM-5.1won 1 · lost 15

GPT-5.1won 1 · lost 16

Gemini 3.6 Flashwon 1 · lost 18

Claude Opus 5.5won 1 · lost 18

GLM-5.2won 1 · lost 19

Qwen3.8 Maxwon 1 · lost 19

Gemini 3.8 Flashwon 1 · lost 19

GPT-5.4won 1 · lost 20

GPT-5.5won 1 · lost 20

Claude Opus 4.7won 1 · lost 20

Kimi K3.1 Prowon 0 · lost 13

Muse Spark 1.3won 0 · lost 16

Muse Wonderwon 0 · lost 16

Claude Sonnet 5.5won 0 · lost 17

GPT-6.1 Solwon 0 · lost 17

Muse Spark 1.2won 0 · lost 17

Gemini 3.7 Flashwon 0 · lost 18

Muse Spark 1.1won 0 · lost 18

Claude Sonnet 5won 0 · lost 19

DeepSeek V4.1 Flashwon 0 · lost 19

GLM-5.3won 0 · lost 19

Kimi K3won 0 · lost 19

Claude Opus 5won 0 · lost 19

Claude Fable 5.1won 0 · lost 19

Claude Fable 5won 0 · lost 19

GPT-5.6 Lunawon 0 · lost 20

Grok 4.6won 0 · lost 20

Grok 4.5won 0 · lost 20

GPT-5.6 Terrawon 0 · lost 20

Claude Opus 4.8won 0 · lost 20

§ 3 · Sources

Where the numbers come from

8 publications, 26 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-10

LMArena Text 1447

GDPval-AA ↗1 measureread 2026-11-10

GDPval-AA 1022

AA-Briefcase ↗1 measureread 2026-11-10

AA-Briefcase 720

ARC-AGI-2 ↗1 measureread 2026-11-10

ARC-AGI-2 18.9%

LiveBench 1.0 ↗8 measuresread 2026-11-10

LiveBench 1.0 66.4LiveBench 1.0 · Reasoning 71.3LiveBench 1.0 · Coding 71.6LiveBench 1.0 · Agentic Coding 41.7LiveBench 1.0 · Mathematics 78.5LiveBench 1.0 · Data Analysis 70.8LiveBench 1.0 · Language 71LiveBench 1.0 · Instruction Following 59.8

Kagi LLM Benchmark ↗1 measureread 2026-11-10

Kagi LLM Benchmark 37.9%

Vals.ai ↗12 measuresread 2026-11-10

Vals · Legal Research Bench 12.5%Vals · CaseLaw 51.66%Vals · Finance Agent 45.36%Vals · CorpFin 60.92%Vals · TaxEval 71.22%Vals · MortgageTax 63.51%Vals · SWE-bench Verified 73% (Mini-SWE-agent)Vals · Vibe Code Bench 47.97% (OpenHands)Vals · Code Migration 12.94%Vals · GPQA Diamond 83.08%Vals · MMLU Pro 84.55%Vals · AIME 95.63%

Vectara Hallucination Leaderboard ↗1 measureread 2026-11-10

Vectara · Factual consistency 94.5%

Badge

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