Skip to main content

OpenAILLM matchup

GPT-5.4vs.GPT-5.4 mini

Is the higher price worth it?

GPT-5.4 performs better than GPT-5.4 mini, but costs 3.3 times as much.

GPT-5.4 ahead
IntelligenceOverall index, 100 points = average
86.4

GPT-5.4

Rank 50

75.1

GPT-5.4 mini

Rank 62

GPT-5.4 mini faster
SpeedTokens per second
64

GPT-5.4

97

GPT-5.4 mini

GPT-5.4 mini cheaper
Price$ per 1M tokens
$5.63

GPT-5.4

$1.69

GPT-5.4 mini

Above 272,000 tokens of context, GPT-5.4 costs more: $5 input and $22.50 output per 1M tokens.

Choose GPT-5.4 for

MultimodalAgentsFinanceCoding

No clear advantage for GPT-5.4 mini.

Where is the difference?

Strengths at a glance

GPT-5.4GPT-5.4 miniMedian of all ranked models

809010070CodingGPT-5.4AgentsGPT-5.4MultimodalGPT-5.4FinanceGPT-5.4
CodingGPT-5.4 ahead
GPT-5.485.0Rank 60
GPT-5.4 mini75.7Rank 71
Median of ranked models87.5

95 % range: GPT-5.4 75.3 to 94.6, GPT-5.4 mini 70.9 to 80.5.

AreaGPT-5.4GPT-5.4 mini

The same values and verdicts as the bars below. In the close-up the net starts at 70 points, just below the weaker model's lowest value; the reference group average is 100. Not in the net because a model has no own value there or the data are too thin: Reasoning, Mathematics, Knowledge, General, Legal, Health.

GPT-5.4 is ahead in 4 of 9 areas, GPT-5.4 mini in none.

GPT-5.4 ahead: 4too little data: 5

Areas in the overall index
MultimodalGPT-5.4 +1488.8 to 75.1
AgentsGPT-5.4 +1290.5 to 78.7
CodingGPT-5.4 +985.0 to 75.7
Reasoningtoo little data87.9 to 72.9
Knowledgetoo little data87.0 to 73.8
Generaltoo little data86.1 to 79.3
Specialist fields
FinanceGPT-5.4 +1185.9 to 75.1
Legaltoo little data83.9 to 71.3
Healthonly GPT-5.4 has a score

Bar length shows the gap in index points (full width 15 points, one standard deviation of the reference group). Solid means clearly ahead, hatched not certain yet, a thin grey bar level.

GPT-5.4 vs. GPT-5.4 mini at the same effort

Both models measured at the same reasoning efforts. This shows whether a lead comes from the model or only from more compute.

xhighGPT-5.4 ahead
9.6 points gap across 4 shared test series
Overall index (xhigh vs. xhigh)GPT-5.4 ahead
GPT-5.4 86.4 vs. GPT-5.4 mini 75.1

The gap compares only the Artificial Analysis test series in which both models are measured at that effort. Tied below 3 points, ahead only beyond two standard errors.

Which model for which task?

Everyday tasks, translated from the area values above. Each row names the values it rests on.

Writing and debugging codeGPT-5.4Coding 85.0 to 75.7
Maths and data analysisopenReasoning 87.9 to 72.9 (too little data)
Facts and researchopenKnowledge 87.0 to 73.8 (too little data)
Agents and workflowsGPT-5.4Agents 90.5 to 78.7
Fast chat repliesGPT-5.4 mini64 to 97 tokens per second
Many requests on a small budgetGPT-5.4 mini$5.63 to $1.69 per 1M tokens
Finance, legal and medical textsGPT-5.4Finance 85.9 to 75.1, Legal 83.9 to 71.3 (too little data), Health without a score for GPT-5.4 mini
Images, screenshots and PDFsGPT-5.4Multimodal 88.8 to 75.1

GPT-5.4 vs. GPT-5.4 mini in our own test

The same prompt for both models. We show the latest attempt without fixes; it does not count toward the overall index.

Pelican on a bicycle

Prompt

Generate an SVG of a pelican riding a bicycle

Task origin: Simon Willison

GPT-5.4

GPT-5.4: Pelican on a bicycle
A community classic for free-form SVG drawing.

Oct 1, 2026 · 117 s · $0.11

Test details
Run date
Oct 1, 2026
API provider
openrouter
Requested API model
openai/gpt-5.4
Reasoning setting
high
Token limit including reasoning
8,192 tokens
Output tokens
7,339 tokens
Cost of the attempt
$0.1101
Response time
116.8 s
API model reported in response
openai/gpt-5.4
Pinned endpoint
OpenAI | openai/gpt-5.4-20260305
Temperature
Provider default
Test protocol
community-visual-tests-v1
Technical checks
Passed, not a quality rating

Visually reviewed. This individual test does not establish overall model quality.

Earlier attempts (1)

  • Sep 7, 2026API request failedReasoning setting: high

GPT-5.4 mini

GPT-5.4 mini: Pelican on a bicycle
A community classic for free-form SVG drawing.

Oct 2, 2026 · 14 s · $0.01

Test details
Run date
Oct 2, 2026
API provider
openrouter
Requested API model
openai/gpt-5.4-mini
Reasoning setting
high
Token limit including reasoning
8,192 tokens
Output tokens
3,142 tokens
Cost of the attempt
$0.0142
Response time
13.5 s
API model reported in response
openai/gpt-5.4-mini
Pinned endpoint
OpenAI | openai/gpt-5.4-mini-20260317
Temperature
Provider default
Test protocol
community-visual-tests-v1
Technical checks
Passed, not a quality rating

Visually reviewed. This individual test does not establish overall model quality.

Earlier attempts (2)

  • Oct 1, 2026API request failedReasoning setting: high
  • Sep 7, 2026API request failedReasoning setting: high
SVG short test

SVG short test

One SVG task with the same fixed settings for every model, without hidden thinking.

Prompt

Return only one standalone SVG of a pelican riding a bicycle. Use viewBox="0 0 800 600", simple shapes, two clearly visible wheels and a recognizable pelican with a long beak. At most 30 SVG elements and 1800 characters. No Markdown, explanations, comments, scripts, animation, CSS or external resources.

GPT-5.4

Unavailable in this round

GPT-5.4 mini

Unavailable in this round

GPT-5.4 and GPT-5.4 mini in the overall index

Every model with a place in the overall index, both models of this matchup highlighted. You can add or hide other models.

Overall index (Gradually Meta-Score v6)

20 of 95 models

Claude Fable 5.2max130.9
Gemini 4 Carbonhigh127.0
Claude Opus 5.5max125.7
Claude Fable 5.1max124.1
Claude Sonnet 5.5max121.9
Gemini 4 Argonhigh121.2
GPT-6 Astramax119.6
Muse Wonder Fullmax118.4
GPT-6.1 Solmax117.5
Claude Opus 5max117.3
Muse Wondermax113.7
Claude Fable 5max112.8
Kimi K3.1 Promax112.6
Muse Ember 1.0max110.5
GPT-5.6 Solmax109.3
GPT-6 Solmax108.4
Gemini 3.7 Flashhigh107.7
DeepSeek V4.1 Flash 1017max107.1
GPT-5.4xhigh86.4
GPT-5.4 minixhigh75.1

All data side by side

The better value is bold. For the overall index only when the gap is clear.

MetricGPT-5.4GPT-5.4 mini
Intelligence
Overall indexpoints, 100 = reference group average, higher is better86.475.1
RankRank 50 of 95Rank 62 of 95
Price per 1M tokens
Inputwhat you send$2.50above 272,000 tokens: $5$0.75
Outputwhat the model writes$15above 272,000 tokens: $22.50$4.50
Cost per task (own tests)average of 1 original-protocol task both models passed (SVG short test not included)$0.11$0.014
Speed
Tokens per second6497
Technical
Context windowhow much text the model reads at once1.05 M tokens≈ 1,580 A4 pages400,000 tokens≈ 600 A4 pages
Knowledge cutoffAug 31, 2025Aug 31, 2025
ReleasedMar 5, 2026Mar 17, 2026
AvailabilityActiveActive
Open weightsnono

Prices without caching or batch discounts · Page count as a rule of thumb from Artificial Analysis

Is there a better choice?

Successor of GPT-5.4GPT-5.5

Overall index 101.4, rank 27$5 input and $30 output per 1M tokens

GPT-5.4 mini vs. GPT-5.5
Cheaper than GPT-5.4 miniMuse Ember 1.0

Overall index 110.5, rank 14$0.10 input and $0.50 output per 1M tokens

View profile
Best model in the overall indexClaude Fable 5.2

Overall index 130.9, rank 1$8 input and $40 output per 1M tokens

Shared benchmarks

Single results from the same benchmark version, task and metric. A value is bold only with the same test setup and a clear gap.

A: GPT-5.4B: GPT-5.4 mini

APEX-Agents-AAOverall · higher is betterGPT-5.4: 33.3%GPT-5.4 mini: 28.2%Artificial AnalysisAs of 11/07/2026
Code MigrationOverall · higher is betterGPT-5.4: 35.0%GPT-5.4 mini: 12.9%Vals AIAs of 10/29/2026
CorpFin v2Overall · higher is betterGPT-5.4: 65.3%GPT-5.4 mini: 60.9%Vals AIAs of 08/12/2026
GDPval-AAOverall · higher is betterGPT-5.4: 1,255.7 eloGPT-5.4 mini: 1,021.8 eloArtificial AnalysisAs of 11/07/2026
GPQA DiamondOverall · higher is betterGPT-5.4: 91.7%GPT-5.4 mini: 83.1%Vals AIAs of 09/01/2026
IFBenchOverall · higher is betterGPT-5.4: 73.9%GPT-5.4 mini: 73.3%Artificial AnalysisAs of 11/07/2026
LiveCodeBenchOverall · higher is betterGPT-5.4: 84.1%GPT-5.4 mini: 81.5%Vals AIAs of 09/01/2026
LMArena Code Arena WebDevOverall · higher is betterTest setup not documented as identicalGPT-5.4: 1,400GPT-5.4 mini: 1,398LMArenaAs of 11/15/2026
LMArena Text, Style ControlOverall · higher is betterGPT-5.4: 1,475GPT-5.4 mini: 1,447LMArenaAs of 11/11/2026
LMArena Vision, Style ControlOverall · higher is betterGPT-5.4: 1,282GPT-5.4 mini: 1,250LMArenaAs of 11/09/2026
Show 54 more benchmarks
MMLU ProOverall · higher is betterGPT-5.4: 87.5%GPT-5.4 mini: 84.6%Vals AIAs of 09/01/2026
MMMU ProOverall · higher is betterGPT-5.4: 87.5%GPT-5.4 mini: 79.2%Vals AIAs of 09/01/2026
MMMU-ProOverall · higher is betterGPT-5.4: 78.4%GPT-5.4 mini: 73.3%Artificial AnalysisAs of 11/07/2026
MortgageTaxOverall · higher is betterGPT-5.4: 68.3%GPT-5.4 mini: 63.5%Vals AIAs of 09/01/2026
ProgramBenchOverall · higher is betterGPT-5.4: 0%GPT-5.4 mini: 0%Vals AIAs of 11/05/2026
SWE-bench VerifiedOverall · higher is betterGPT-5.4: 78.2%GPT-5.4 mini: 73.0%Vals AIAs of 09/01/2026
TaxEval v2Overall · higher is betterGPT-5.4: 74.0%GPT-5.4 mini: 71.2%Vals AIAs of 09/01/2026
Vibe Code Bench v1.1Overall · higher is betterGPT-5.4: 67.4%GPT-5.4 mini: 48.0%Vals AIAs of 11/05/2026
Code MigrationCLI · higher is betterGPT-5.4: 23.8%GPT-5.4 mini: 8.2%Vals AIAs of 10/29/2026
CorpFin v2Exact Pages · higher is betterGPT-5.4: 64.1%GPT-5.4 mini: 61.5%Vals AIAs of 08/12/2026
GPQA DiamondFew-Shot CoT · higher is betterGPT-5.4: 91.4%GPT-5.4 mini: 84.8%Vals AIAs of 09/01/2026
LiveCodeBenchEasy · higher is betterGPT-5.4: 99.1%GPT-5.4 mini: 97.8%Vals AIAs of 09/01/2026
LMArena Code Arena WebDevHTML development · higher is betterTest setup not documented as identicalGPT-5.4: 1,423GPT-5.4 mini: 1,420LMArenaAs of 11/15/2026
LMArena Text, Style ControlBusiness, management and finance · higher is betterGPT-5.4: 1,482GPT-5.4 mini: 1,459LMArenaAs of 11/11/2026
LMArena Vision, Style ControlChinese · higher is betterGPT-5.4: 1,317GPT-5.4 mini: 1,285LMArenaAs of 11/09/2026
MMLU ProBiology · higher is betterGPT-5.4: 92.9%GPT-5.4 mini: 91.4%Vals AIAs of 09/01/2026
MortgageTaxNumerical Extraction · higher is betterGPT-5.4: 76.0%GPT-5.4 mini: 79.4%Vals AIAs of 09/01/2026
ProgramBenchAlmost Resolved · higher is betterGPT-5.4: 7%GPT-5.4 mini: 3%Vals AIAs of 11/05/2026
SWE-bench Verified<15 min fix · higher is betterGPT-5.4: 88.1%GPT-5.4 mini: 85.6%Vals AIAs of 09/01/2026
TaxEval v2Correctness · higher is betterGPT-5.4: 64.9%GPT-5.4 mini: 63.0%Vals AIAs of 09/01/2026
Code MigrationCOBOL · higher is betterGPT-5.4: 68.6%GPT-5.4 mini: 27.1%Vals AIAs of 10/29/2026
CorpFin v2Max Fitting Context · higher is betterGPT-5.4: 64.9%GPT-5.4 mini: 61.0%Vals AIAs of 08/12/2026
GPQA DiamondZero-Shot CoT · higher is betterGPT-5.4: 91.9%GPT-5.4 mini: 81.3%Vals AIAs of 09/01/2026
LiveCodeBenchHard · higher is betterGPT-5.4: 67.7%GPT-5.4 mini: 61.7%Vals AIAs of 09/01/2026
LMArena Code Arena WebDevReact development · higher is betterTest setup not documented as identicalGPT-5.4: 1,388GPT-5.4 mini: 1,386LMArenaAs of 11/15/2026
LMArena Text, Style ControlChinese · higher is betterGPT-5.4: 1,505GPT-5.4 mini: 1,473LMArenaAs of 11/11/2026
LMArena Vision, Style ControlCreative writing with images · higher is betterGPT-5.4: 1,264GPT-5.4 mini: 1,225LMArenaAs of 11/09/2026
MMLU ProBusiness · higher is betterGPT-5.4: 91.8%GPT-5.4 mini: 89.7%Vals AIAs of 09/01/2026
MortgageTaxSemantic Extraction · higher is betterGPT-5.4: 60.7%GPT-5.4 mini: 47.6%Vals AIAs of 09/01/2026
ProgramBenchFully Resolved · higher is betterGPT-5.4: 0%GPT-5.4 mini: 0%Vals AIAs of 11/05/2026
SWE-bench Verified>4 hours · higher is betterGPT-5.4: 0%GPT-5.4 mini: 0%Vals AIAs of 09/01/2026
TaxEval v2Stepwise Reasoning · higher is betterGPT-5.4: 83.0%GPT-5.4 mini: 79.4%Vals AIAs of 09/01/2026
Code MigrationCode Quality · higher is betterGPT-5.4: 3.632 / 5GPT-5.4 mini: 3.053 / 5Vals AIAs of 10/29/2026
CorpFin v2Shared Max Context · higher is betterGPT-5.4: 66.8%GPT-5.4 mini: 60.3%Vals AIAs of 08/12/2026
LiveCodeBenchMedium · higher is betterGPT-5.4: 85.6%GPT-5.4 mini: 84.9%Vals AIAs of 09/01/2026
LMArena Code Arena WebDevWeb development · higher is betterTest setup not documented as identicalGPT-5.4: 1,400GPT-5.4 mini: 1,398LMArenaAs of 11/15/2026
LMArena Text, Style ControlCoding · higher is betterGPT-5.4: 1,521GPT-5.4 mini: 1,495LMArenaAs of 11/11/2026
LMArena Vision, Style ControlDiagram understanding · higher is betterGPT-5.4: 1,319GPT-5.4 mini: 1,283LMArenaAs of 11/09/2026
MMLU ProChemistry · higher is betterGPT-5.4: 91.1%GPT-5.4 mini: 89.1%Vals AIAs of 09/01/2026
ProgramBenchRaw Pass Rate · higher is betterGPT-5.4: 68.6%GPT-5.4 mini: 54.4%Vals AIAs of 11/05/2026
SWE-bench Verified1-4 hours · higher is betterGPT-5.4: 50.0%GPT-5.4 mini: 33.3%Vals AIAs of 09/01/2026
LMArena Text, Style ControlCreative writing · higher is betterGPT-5.4: 1,440GPT-5.4 mini: 1,401LMArenaAs of 11/11/2026
LMArena Vision, Style ControlEnglish · higher is betterGPT-5.4: 1,279GPT-5.4 mini: 1,247LMArenaAs of 11/09/2026
MMLU ProComputer Science · higher is betterGPT-5.4: 89.8%GPT-5.4 mini: 87.1%Vals AIAs of 09/01/2026
SWE-bench Verified15 min - 1 hour · higher is betterGPT-5.4: 76.2%GPT-5.4 mini: 70.9%Vals AIAs of 09/01/2026
LMArena Text, Style ControlEnglish · higher is betterGPT-5.4: 1,476GPT-5.4 mini: 1,450LMArenaAs of 11/11/2026
LMArena Vision, Style ControlEntity recognition · higher is betterGPT-5.4: 1,192GPT-5.4 mini: 1,203LMArenaAs of 11/09/2026
MMLU ProEconomics · higher is betterGPT-5.4: 89.7%GPT-5.4 mini: 88.7%Vals AIAs of 09/01/2026
LMArena Text, Style ControlEntertainment, sports and media · higher is betterGPT-5.4: 1,441GPT-5.4 mini: 1,408LMArenaAs of 11/11/2026
LMArena Vision, Style ControlHomework · higher is betterGPT-5.4: 1,330GPT-5.4 mini: 1,291LMArenaAs of 11/09/2026
MMLU ProEngineering · higher is betterGPT-5.4: 85.2%GPT-5.4 mini: 82.0%Vals AIAs of 09/01/2026
LMArena Text, Style ControlExcluding ties · higher is betterGPT-5.4: 1,484GPT-5.4 mini: 1,449LMArenaAs of 11/11/2026
LMArena Vision, Style ControlHumor · higher is betterGPT-5.4: 1,267GPT-5.4 mini: 1,239LMArenaAs of 11/09/2026
MMLU ProHealth · higher is betterGPT-5.4: 82.4%GPT-5.4 mini: 80.1%Vals AIAs of 09/01/2026

More comparisons

Compare GPT-5.4 and GPT-5.4 mini with the models people compare them with most often.

Compare another matchup