The LLM benchmark aggregator.

GPT-5.4 Nano

OpenAI

Strongest in Factual grounding (#2 of 101), weakest in Core abilities (#192 of 204). Above par in 9 of 24 scopes. Among the models it meets almost everywhere, it finishes behind Claude Fable 5 and Claude Fable 5.1 and ahead of Grok 4.3 and DeepSeek V3.

§ 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.

Safety55.4−6.1#43/3371/3

Factual grounding67.3−3.3#2/1011/1

Reasoning50.4−16.5#88/1783/4

Mathematics55.5−10.2#41/1402/2

Science51−12.5#74/1221/1

Reasoning43.9−25.2#84/1402/3

Knowledge45.8−17.7#106/1381/2

Academic knowledge44.9−19#98/1231/1

Agents50.1−18#117/2682/5

Knowledge work50.2−23.4#83/1782/2

Human preference57.1−10.4#123/3421/1

Professional45.2−16.4#129/1681/1

Medical48.9−17.9#89/1401/1

Finance47.6−15.3#101/1511/1

Legal39.9−27.2#130/1511/1

Coding42.3−26.8#141/1652/5

Code generation36.5−33.5#66/771/2

Agentic coding44−24.7#114/1572/4

Core abilities39.3−28.1#192/2042/3

Instruction following48.8−24.9#31/571/1

Data analysis40.2−25.9#48/571/1

Language26−45.5#57/571/1

General intelligence40.6−29.5#169/2042/3

§ 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 Nano, left for the other.

GPT-4o Miniwon 18 · lost 1

GPT-4.1 Nanowon 17 · lost 2

Mistral Smallwon 13 · lost 2

Llama 4 Scout Instructwon 15 · lost 3

Command Awon 14 · lost 4

GPT-4owon 13 · lost 4

GPT-5 Nanowon 13 · lost 4

Mistral Medium 3.5won 13 · lost 4

GPT OSS 20Bwon 12 · lost 4

Claude 3.5 Haikuwon 11 · lost 4

Llama 4 Maverick Instructwon 14 · lost 6

Gemini 2.5 Flash Litewon 10 · lost 6

O3 Miniwon 10 · lost 8

Claude Haiku 4.5won 11 · lost 9

DeepSeek V3won 11 · lost 10

Grok 4.3won 12 · lost 11

Qwen3.6 27Bwon 8 · lost 8

GPT OSS 120Bwon 10 · lost 10

Mistral Large 3won 10 · lost 10

GPT-4.1 Miniwon 9 · lost 10

Gemini 2.0 Flashwon 8 · lost 9

O4 Miniwon 9 · lost 11

Claude 3.7 Sonnetwon 7 · lost 9

Gemini 3.1 Flash Litewon 8 · lost 11

Gemma 4 31Bwon 7 · lost 10

GPT-5.4 Miniwon 9 · lost 14

Grok 3won 6 · lost 10

Kimi K2won 7 · lost 12

Gemini 3.5 Flash Litewon 8 · lost 14

DeepSeek R1won 7 · lost 13

Claude 3.5 Sonnetwon 5 · lost 10

Qwen3.5 Flashwon 5 · lost 10

Claude Sonnet 4won 7 · lost 14

Qwen3.6 Pluswon 7 · lost 15

Qwen3 235B A22B Instructwon 6 · lost 13

Nemotron 3 Ultrawon 5 · lost 11

GPT-5 Miniwon 6 · lost 14

Gemini 2.5 Flashwon 6 · lost 14

DeepSeek V3.2won 6 · lost 14

Grok 4.1 Fastwon 5 · lost 12

Kimi K2.6won 7 · lost 17

Grok 4 Fastwon 5 · lost 13

GPT-4.1won 5 · lost 14

GLM-4.6won 5 · lost 14

MiniMax M3won 6 · lost 17

Grok 3 Miniwon 4 · lost 12

Gemini 2.5 Prowon 5 · lost 16

Mimo V2.5 Prowon 4 · lost 14

GLM-5.3 Flashwon 5 · lost 18

Claude Opus 4won 4 · lost 15

Gemini 3.1 Prowon 5 · lost 19

O1won 3 · lost 12

Qwen3 Maxwon 3 · lost 12

Mimo V2.5won 3 · lost 12

MiniMax M2.5won 3 · lost 13

MiniMax M2.7won 3 · lost 13

Gemini 3 Flashwon 3 · lost 13

Claude Opus 4.6won 4 · lost 18

Claude Opus 4.5won 4 · lost 19

GLM-5.2won 4 · lost 19

Claude Sonnet 4.6won 4 · lost 20

GPT-6 Lunawon 3 · lost 17

Kimi K2.5won 3 · lost 17

Qwen3.8 27Bwon 3 · lost 19

Kimi K2 Thinkingwon 2 · lost 13

DeepSeek V4 Flashwon 3 · lost 20

GPT-5.6 Lunawon 3 · lost 20

Grok 4.5won 3 · lost 20

Qwen3.7 Maxwon 3 · lost 20

Gemini 3.5 Flashwon 3 · lost 20

Gemini 3.6 Flashwon 3 · lost 20

Gemini 3.8 Flashwon 3 · lost 20

Inkling Smallwon 2 · lost 14

GLM-4.7won 2 · lost 14

DeepSeek V4 Prowon 3 · lost 21

GLM-5.1won 2 · lost 15

Grok 4won 2 · lost 16

GLM-5won 2 · lost 16

Gemini 3 Prowon 2 · lost 17

Grok 4.7won 2 · lost 18

GPT-6 Solwon 2 · lost 19

DeepSeek V4.1 Flashwon 2 · lost 19

GPT-6 Astrawon 2 · lost 19

GPT-5.2won 2 · lost 21

Inklingwon 2 · lost 21

Qwen3.8 Maxwon 2 · lost 21

GPT-5.4won 2 · lost 22

GPT-5.5won 2 · lost 22

GPT-5.6 Solwon 2 · lost 22

Claude Opus 4.7won 2 · lost 22

GPT-5.1won 1 · lost 17

O3won 1 · lost 18

GPT-5won 1 · lost 19

Claude Sonnet 4.5won 1 · lost 19

Claude Opus 5.5won 1 · lost 19

Muse Spark 1.2won 1 · lost 20

GLM-5.3won 1 · lost 21

Gemini 3.7 Flashwon 1 · lost 21

Kimi K3won 1 · lost 21

Grok 4.6won 1 · lost 22

Claude Sonnet 5won 1 · lost 22

GPT-5.6 Terrawon 1 · lost 22

Claude Opus 4.8won 1 · lost 22

Claude Opus 5won 1 · lost 22

Muse Spark 1.1won 1 · lost 22

Muse Sparkwon 0 · lost 15

Muse Spark 1.3won 0 · lost 18

Claude Fable 5.1won 0 · lost 22

Claude Fable 5won 0 · lost 22

§ 3 · Sources

Where the numbers come from

8 publications, 30 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-09-26

LMArena Text 1401

GDPval-AA ↗1 measureread 2026-09-26

GDPval-AA 937

AA-Briefcase ↗1 measureread 2026-09-26

AA-Briefcase 671

ARC-AGI-2 ↗1 measureread 2026-09-26

ARC-AGI-2 5.7%

LiveBench ↗8 measuresread 2026-09-26

LiveBench 69.6LiveBench · Reasoning 81.1LiveBench · Coding 70.8LiveBench · Agentic Coding 46.8LiveBench · Mathematics 91LiveBench · Data Analysis 67.6LiveBench · Language 62.5LiveBench · Instruction Following 67.2

Kagi LLM Benchmark ↗1 measureread 2026-09-26

Kagi LLM Benchmark 39.7%

Vals.ai ↗16 measuresread 2026-09-26

Vals · Legal Research Bench 6.25%Vals · CaseLaw 51.88%Vals · LegalBench 77.92%Vals · Harvey Legal Agent Benchmark 0%Vals · Finance Agent 38.22%Vals · CorpFin 61.19%Vals · TaxEval 67.42%Vals · MortgageTax 59.1%Vals · MedCode 41.03%Vals · MedScribe 77.09%Vals · SWE-bench Verified 69.8% (Mini-SWE-agent)Vals · Vibe Code Bench 26.1% (OpenHands)Vals · Code Migration 14.47%Vals · GPQA Diamond 77.53%Vals · MMLU Pro 77.17%Vals · AIME 88.75%

Vectara Hallucination Leaderboard ↗1 measureread 2026-09-26

Vectara · Factual consistency 96.9%

Badge

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