Noometry

OpenAI, proprietary

GPT-5.4 mini

GPT-5.4 mini by OpenAI ranks 76th of 354 ranked models on the Noometry Index as of October 2026, with a score of 45.0. Its strongest category is multimodal, where it ranks 56th. API pricing starts at $0.75 per million input tokens and $4.50 per million output tokens, with a 400K-token context window.

Last verified

Noometry rank
#76 of 354
Index score
45.0
Evidence
Confirmed 46 results
Provider
OpenAI
Released
March 17, 2026
Weights
Proprietary
Reasoning
Yes
Context window
400K
Max output
128K
Input price
$0.75 / M
Output price
$4.50 / M
Blended price
$1.69 / M
Output speed
10 tokens/s Kagi
Value
#140 of 219
Knowledge cutoff
August 2025
Input
text, image

Category scores

Each category score combines every public result we have in that category.

  1. Coding 45.2
  2. Agentic & Tool Use 29.9
  3. Reasoning 30.4
  4. Math 45.5
  5. Knowledge 51.5
  6. Multimodal 39.7
  7. Multilingual 51.9
  8. Instruction Following 74.1
  9. Long Context 43.0
  10. Writing & Preference 64.0
CategoryScoreRankResults
Coding45.2#725
Agentic & Tool Use29.9#811
Reasoning30.4#8511
Math45.5#755
Knowledge51.5#674
Multimodal39.7#561
Multilingual51.9#961
Instruction Following74.1#1021
Long Context43.0#1121
Writing & Preference64.0#584

Strengths and weaknesses

Categories where GPT-5.4 mini places highest and lowest among the models ranked in each, with its score against that category's median.

Strongest categories

CategoryScorevs medianRank
Writing & Preference64.0+10.3#58 of 312, top 19%
Coding45.2+6.5#72 of 340, top 22%
Knowledge51.5+14.2#67 of 314, top 22%

Weakest categories

CategoryScorevs medianRank
Agentic & Tool Use29.9−0.4#81 of 154, top 53%
Multimodal39.7+1.1#56 of 128, top 44%
Long Context43.0+2.0#112 of 296, top 38%

Closest competitors

The models ranked just above and below GPT-5.4 mini. When scores are this close, price and speed are often the better way to choose.

ModelRankScoreBlended $/MSpeed
Qwen3.7 Plus#7245.3$0.70—Compare
Hy4 preview#7345.3$1.13—Compare
MiMo-V2.5-Pro#7445.2$0.54—Compare
Gemini 2.5 Pro#7545.0$3.445Compare
Amazon Nova Experimental Chat 26 02 10#7744.5——Compare
DeepSeek-V3.2-Exp#7844.3$0.2916Compare
Hy3#7944.2$0.14—Compare
Inkling#8044.1$2.57—Compare

Sponsored placements are available on pages like this one. Advertise on Noometry

Benchmark results

Every published result we track, with its source. Bold rows are the ones used for ranking; where several exist we prefer independent runs over self-reported numbers.

Coding

BenchmarkScorePositionSettingSourceDate
FrontierCode27%#28 of 37, top 76%Epoch AI
LMArena WebDev1397#74 of 113, top 66%highLMArena2026-10-08
SciCode49.9%#45 of 121, top 38%xhighEpoch AI
WeirdML60.3%#32 of 119, top 27%highEpoch AI
WeirdML37.9%noneEpoch AI
LMArena Coding1438#100 of 294, top 35%highLMArena2026-10-08
ALE-Bench1,189#32 of 105, top 31%highEpoch AI

Agentic & Tool Use

BenchmarkScorePositionSettingSourceDate
DeepResearch Bench36.3%#22 of 24, top 92%lowEpoch AI

Reasoning

BenchmarkScorePositionSettingSourceDate
ARC-AGI-213.2%highEpoch AI
ARC-AGI-21.1%lowEpoch AI
ARC-AGI-24.4%mediumEpoch AI
ARC-AGI-218.9%#42 of 83, top 51%xhighEpoch AI
Kagi LLM Benchmark37.9%#81 of 99, top 82%Kagi LLM Benchmark
NYT Connections (extended)61.8%#55 of 91, top 61%xhigh reasoningLech Mazur benchmarks
ARC-AGI-158%highEpoch AI
ARC-AGI-113%lowEpoch AI
ARC-AGI-140.8%mediumEpoch AI
ARC-AGI-163.7%#48 of 83, top 58%xhighEpoch AI
CritPt10%#47 of 134, top 36%xhighEpoch AI
Chess Puzzles18%highEpoch AI2026-04-15
Chess Puzzles3%noneEpoch AI2026-08-07
Chess Puzzles24%#45 of 129, top 35%xhighEpoch AI2026-08-07
Thematic Generalization61.7%#12 of 23, top 53%xhigh reasoningLech Mazur benchmarks
LMArena Hard Prompts1424#98 of 297, top 33%highLMArena2026-10-08
Mystery Game Puzzles8%lowEpoch AI2026-08-27
Mystery Game Puzzles7%mediumEpoch AI2026-08-27
Mystery Game Puzzles11%#59 of 74, top 80%noneEpoch AI2026-08-27
DTBench80%#76 of 151, top 51%xhighEpoch AI
LMCA40.8%#44 of 125, top 36%xhighEpoch AI
Epoch Capabilities Index148.84#60 of 213, top 29%Epoch AI2026-03-17
ForecastBench57#63 of 72, top 88%Epoch AI

Math

BenchmarkScorePositionSettingSourceDate
FrontierMath (Tiers 1-3)24.6%lowEpoch AI2026-08-28
FrontierMath (Tiers 1-3)17.2%noneEpoch AI2026-08-28
FrontierMath (Tiers 1-3)51.2%#47 of 81, top 59%xhighEpoch AI2026-06-12
FrontierMath Tier 49.8%#53 of 63, top 85%xhighEpoch AI2026-06-12
OTIS Mock AIME 2024-202587.2%highEpoch AI2026-04-15
OTIS Mock AIME 2024-202526.7%noneEpoch AI2026-08-07
OTIS Mock AIME 2024-202588.9%#54 of 173, top 32%xhighEpoch AI2026-08-07
ProofBench21%#48 of 77, top 63%xhighEpoch AI
LMArena Math1419#105 of 285, top 37%highLMArena2026-10-08
FrontierMath (Feb 2025 set)28.3%#20 of 68, top 30%highEpoch AI2026-04-15
FrontierMath Tier 4 (v1)2.1%#44 of 55, top 80%highEpoch AI2026-04-15

Knowledge

BenchmarkScorePositionSettingSourceDate
GPQA Diamond83.6%highEpoch AI2026-04-15
GPQA Diamond64.1%noneEpoch AI2026-08-07
GPQA Diamond86.9%#57 of 186, top 31%xhighEpoch AI2026-08-07
SimpleQA Verified29.4%#59 of 77, top 77%highEpoch AI2026-08-27
Vectara Hallucination Rate (lower is better)5.5%#17 of 96, top 18%Vectara Hallucination Leaderboard
LMArena Expert1435#90 of 273, top 33%highLMArena2026-10-08

Multimodal

BenchmarkScorePositionSettingSourceDate
LMArena Vision1245#59 of 122, top 49%highLMArena2026-10-09

Multilingual

BenchmarkScorePositionSettingSourceDate
LMArena Non-English1405#97 of 297, top 33%highLMArena2026-10-08
LMArena Chinese1446#108 of 285, top 38%highLMArena2026-10-08
LMArena French1440#83 of 223, top 38%highLMArena2026-10-08
LMArena German1409#87 of 231, top 38%highLMArena2026-10-08
LMArena Japanese1374#84 of 211, top 40%highLMArena2026-10-08
LMArena Korean1368#88 of 213, top 42%highLMArena2026-10-08
LMArena Russian1417#84 of 283, top 30%highLMArena2026-10-08
LMArena Spanish1405#105 of 226, top 47%highLMArena2026-10-08

Instruction Following

BenchmarkScorePositionSettingSourceDate
LMArena Instruction Following1405#94 of 298, top 32%highLMArena2026-10-08

Long Context

BenchmarkScorePositionSettingSourceDate
LMArena Longer Query1407#110 of 291, top 38%highLMArena2026-10-08

Writing & Preference

BenchmarkScorePositionSettingSourceDate
LMArena Text1412#107 of 297, top 37%highLMArena2026-10-08
LMArena Creative Writing1370#110 of 295, top 38%highLMArena2026-10-08
EQ-Bench Creative Writing1665#38 of 115, top 34%EQ-Bench
LMArena Multi-Turn1429#84 of 295, top 29%highLMArena2026-10-08

API pricing by provider

RouteInput $/MOutput $/MCached input $/MChecked
azure$0.75$4.50$0.0752026-10-10
openai$0.75$4.50$0.0752026-10-10
openrouter$0.75$4.50$0.0752026-10-10

All OpenAI API prices → Estimate your cost →

Compare GPT-5.4 mini

Other OpenAI models

Frequently asked questions

How good is GPT-5.4 mini?

GPT-5.4 mini by OpenAI ranks 76th of 354 ranked models on the Noometry Index as of October 2026, with a score of 45.0. Its strongest category is multimodal, where it ranks 56th. API pricing starts at $0.75 per million input tokens and $4.50 per million output tokens, with a 400K-token context window.

How much does GPT-5.4 mini cost?

GPT-5.4 mini costs $0.75 per million input tokens and $4.50 per million output tokens on OpenAI's own API, with cached input at $0.075.

What is GPT-5.4 mini's context window?

GPT-5.4 mini accepts up to 400K tokens of input and can write up to 128K tokens in one response.

Is GPT-5.4 mini open source?

No. GPT-5.4 mini is proprietary and available only through OpenAI's API and partner platforms.

How fast is GPT-5.4 mini?

GPT-5.4 mini generated about 10 output tokens per second in the Kagi LLM Benchmark's timed runs. Speed varies by provider, load and reasoning effort.

What are GPT-5.4 mini's strengths and weaknesses?

Relative to other ranked models, GPT-5.4 mini places best in writing & preference, coding, knowledge and lowest in agentic & tool use, multimodal, long context.

What is GPT-5.4 mini best at?

Its best category is multimodal, where it ranks 56th on Noometry.