OpenAIUnited States· released Mar 2026

GPT-5.4 mini

Name: GPT-5.4 mini
Author: OpenAI

Ranked #34, 50.7% behind the frontier on average.

OpenAI, United States. One of 26 models from this lab on the Index.

Third-party publishedMedium confidence (7+ comparable tests, 3+ Index categories)

Snapshot September 8, 2026

OpenCharts Index

49.3

Ranked #34 (33 to 34) of 36 · 50.7% behind the frontier on average · 17 comparable tests · 3 counting categories

Drop any single test and the Index lands between 44.3 and 53.0, anywhere from #33 to #34. Neighbours inside that range are ties.

Needs more results (3 of 8 so far)

Averaging 28.2 on the 3 it has; the score opens at 4 of 8.

Distance to today's frontier on the hardest tests. Not a measure of intelligence.

What it's known for

GPT-5.4 mini, in 17 results.

Strongest in coding (53.5) and reasoning (50.7). It trails the frontier most in math, 56.3% behind on average. No comparable result yet in long context. Between 50% and 60% behind the best published results, on average. Drop any single test and it would sit anywhere from #33 to #34.

Strongest category

Coding

Writing, fixing and shipping real code, judged by tests or users.

53.5

SciCode49.9% · 80% of best
WeirdML60.3% · 65% of best
FrontierCode27.0% · 51% of best
Code Arena (WebDev)1,397 · 18% win rate vs leader

Mean of 4 of the category's 7 comparable tests. Counts toward the Index.

Best single result

Mock AIME 2024–2025

Solve competition mathematics problems that have a single exact integer answer.

88.9%

88.9% of the best (100.0%) · #41 of 78.

Olympiad-style problems need long, careful chains of reasoning, and there is no partial credit.

Where it trails

FrontierMath Tier 4

Solve the hardest FrontierMath tier: problems that take expert mathematicians days.

9.8%

90.0% behind the best published result (97.6%) · #43 of 51.

The deepest end of the set. Progress here signals genuinely new capability.

Not measured yet

No comparable result yet in this category. A blank is a blank, never a zero, and it does not lower the Index.

Long context

Covered, but not averaged into the Index:

Knowledgesingle-test category, shown beside the Index · 38.9
Agentic2 of 8 tests · needs 4 · mean 58.8
Multimodalsingle-test category, shown beside the Index · 82.6
Human preferencesingle-test category, shown beside the Index · 83.2

Every score, with its source

See it at work

What GPT-5.4 mini was asked to do, and how it did.

Every test it has taken, by category, with the kind of task it faced, who produced the number and where the result landed against the best published one. The examples are original and illustrative, never items from the datasets themselves.

Human preference · mean 83.2 across 1 comparable test · a single-test category, shown beside the Index and never averaged into it

Text ArenaLeaderboard

Answer the same prompt as a rival model; real users vote blind on which answer they prefer.

1,448

42% expected win rate against the board leader (1,507)
#51 of 92

high · board 2026-09-02

Source: Arena

Every result is read against the best published one on its test: percent scores as a share of the best above chance, Arena ratings as the expected win rate against the board leader, open-ended values on a log scale. Sitting a harder exam never lowers a model. Every number links to the publisher that produced it.

Head to head

GPT-5.4 mini against whoever you pick.

Choose a rival. Every counted test both have taken appears side by side as a share of the best published result on that test, with the category means above. A win is a higher share; a gap under 2 points is a tie, because one item on a 45-problem exam is 2.2 points and publishers' own standard errors are of that order.

GPT-5.4 mini

#34·Index 49.3D

0 : 14

wins · 0 ties · 14 shared

GPT-6 Astra

#1·Index 99.0A+

GPT-6 Astra comes out ahead on 14 of the 14 tests both have taken.

Category means

Reasoning
50.7
99.6
Knowledge
38.9
100.0
Coding
53.5
97.7
Math
43.7
99.8
Agentic
58.8
98.5
Multimodal
82.6
—
Human preference
83.2
—

Shared tests · biggest gaps first

FrontierMath Tier 4Math
9.8%
97.6%
−90.0
Code Arena (WebDev)Coding
1,397
1,797
−81.8
ARC-AGI-2Reasoning
18.9%
95.0%
−80.1
ProofBenchMath
21.0%
99.0%
−78.0
CritPtReasoning
10.0%
31.7%
−67.2
SimpleQA VerifiedKnowledge
29.4%
75.6%
−61.1
FrontierCodeCoding
27.0%
53.3%
−49.0
APEX-AgentsAgentic
24.6%
46.7%
−46.6
FrontierMath (Tiers 1–3)Math
51.2%
93.7%
−45.3
ARC-AGI-1Reasoning
63.7%
98.5%
−35.4
WeirdMLCoding
60.3%
92.9%
−35.1
GPQA DiamondReasoning
86.9%
95.8%
−12.6
Mock AIME 2024–2025Math
88.9%
100.0%
−11.1
SciCodeCoding
49.9%
56.5%
−10.6

Bars are distances to the best published result on each test (percent scores above chance, Arena ratings as win rate against the leader, open-ended values on a log scale). Raw values as the publishers report them.

Open GPT-6 Astra

Scores

Every test, every source

The best published run per benchmark, the raw value as the publisher reports it, who produced the number, how close it comes to the best published result on that test, and where the model sits among every model scored on it. Percent scores are read above chance, Arena ratings as a win rate against the board leader, open-ended values on a log scale. A test with too few models to compare against is shown but not counted.

Category means

Reasoning50.7 · 4/6
Knowledgebeside38.9 · 1/1
Coding53.5 · 4/7
Math43.7 · 4/5
Agentic58.8 · 2/8
Multimodalbeside82.6 · 1/1
Human preferencebeside83.2 · 1/1
Long context—

Mean · tests taken of the category's comparable tests. A faded category either rests on fewer than half of them or is a single-test category (marked “beside”), which is shown next to the Index and never averaged into it.

GPQA DiamondEpoch-runReasoning

86.9%

87.4% of the best (95.8%), above chance
40th percentile of 78 · n = 198

2026-03-17 · xhigh

Source: Epoch AI — AI Benchmarking Hub

ARC-AGI-2LeaderboardReasoning

18.9%

19.9% of the best (95.0%)
32th percentile of 51 · n = 120

2026-03-17 · xhigh

Source: Epoch AI — AI Benchmarking Hub

ARC-AGI-1LeaderboardReasoning

63.7%