GPT-5.4 vs GPT-5.4 mini

Change models
GPT-5.4OpenAI
IndexAll categories, out of 100.
64.8 ±2.5 best
54.5 ±2.4
GapIndex points below the top.
Top
−10.3
CoverageIndex weight that has results.
95%
100% best
SpeedOutput tokens per second.
38/s
95/s best
EloLMArena rating and rank.
1452 best
1412
ReleasedMonth of release.
2026-03
2026-03
Input / 1MUS dollars per 1M input tokens.
$2.50
$0.75 best
Output / 1MUS dollars per 1M output tokens.
$15.00
$4.50 best
Batch output / 1MBatch price per 1M output tokens.
$7.50
$2.25 best
1K chats2K tokens in, 500 out, each.
$12.50
$3.75 best
1K documents20K tokens in, 1K out, each.
$65.00
$19.50 best
1K agent tasks200K tokens in, 20K out, each.
$800.00
$240.00 best
ContextMaximum tokens in one request.
1.05M best
400K
AgenticMulti-step tasks with tools.
62.1 best
56.1
CodingCode writing and repair.
61.8 best
55.1
ReasoningLogic problems and puzzles.
69.0 best
52.2
MultimodalTasks with images and text.
62.1 best
54.8
KnowledgeFacts and expert knowledge.
66.4 best
53.7
MultilingualTasks in many languages.
—
57.6
InstructionTasks with strict prompt rules.
63.4 best
52.6
MathMath problems.
68.0 best
53.9

Benchmarks

57 shared
Only GPT-5.4: 42Only GPT-5.4 mini: 18
GPT-5.4
GPT-5.4 mini
43.7% best31 / 222
28.1%91 / 222
82.0% best29 / 217
77.0%82 / 217
93.3% best19 / 217
86.9%65 / 217
23.4% best23 / 214
10.0%65 / 214
95.3% best46 / 200
88.9%63 / 200
92.0% best22 / 199
87.5%56 / 199
50.8% best18 / 164
37.5%63 / 164
98.9% best2 / 149
93.4%31 / 149
73.9% best26 / 146
73.3%29 / 146
TaxEval v2Finance
74.0% best42 / 143
71.2%80 / 143
84.1% best40 / 142
81.5%60 / 142
92.8% best13 / 141
88.0%38 / 141
91.7% best18 / 135
83.1%57 / 135
MMLU ProKnowledge
87.5% best26 / 135
84.6%60 / 135
37.3% best47 / 133
25.6%71 / 133
74.0% best24 / 131
18.9%61 / 131
CorpFin v2Finance
65.3% best28 / 131
60.9%62 / 131
93.7% best26 / 126
63.7%64 / 126
48.5% best69 / 123
48.0%70 / 123
84.2% best20 / 122
17.8%55 / 122