Noometry

Model comparison

GPT-5.4 mini vs Step 3

GPT-5.4 mini is the stronger model overall, scoring 45.0 to 40.5 on the Noometry Index.

Last verified . 17 shared benchmarks.

GPT-5.4 mini OpenAI

45.0

Rank #76 Confirmed

Step 3 StepFun

40.5

Rank #149 Confirmed

Summary

  • They share 17 benchmarks with published results for both. GPT-5.4 mini scores higher in 9 categories and Step 3 in 0 categories; 9 gaps are clear of the uncertainty.
  • The widest gap is in knowledge, where GPT-5.4 mini leads 51.5 to 36.8.
  • The biggest single-benchmark swing is Kagi LLM Benchmark: 37.9% for GPT-5.4 mini and 62.3% for Step 3.
  • Step 3 has downloadable open weights; the other is API-only.

Side by side

GPT-5.4 miniStep 3
ProviderOpenAIStepFun
Noometry Index45.040.5
Released2026-03-17—
WeightsProprietaryOpen
Context window400K—
Max output128K—
Input $ / M tokens$0.75—
Output $ / M tokens$4.50—
Results tracked4617

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

Category by category

Coding GPT-5.4 mini leads

GPT-5.4 mini: 45.2 (#72), Step 3: 40.1 (#147)

BenchmarkGPT-5.4 miniStep 3
LMArena Coding14381367
FrontierCode27%—
LMArena WebDev1397—
SciCode49.9%—
WeirdML60.3%—
ALE-Bench1,189—

Agentic & Tool Use Not comparable

GPT-5.4 mini: 29.9 (#81), Step 3: —

BenchmarkGPT-5.4 miniStep 3
DeepResearch Bench36.3%—

Reasoning GPT-5.4 mini leads

GPT-5.4 mini: 30.4 (#85), Step 3: 28.4 (#105)

BenchmarkGPT-5.4 miniStep 3
Kagi LLM Benchmark37.9%62.3%
LMArena Hard Prompts14241355
ARC-AGI-218.9%—
NYT Connections (extended)61.8%—
ARC-AGI-163.7%—
CritPt10%—
Chess Puzzles24%—
Thematic Generalization61.7%—
Mystery Game Puzzles11%—
DTBench80%—
LMCA40.8%—
Epoch Capabilities Index148.84—
ForecastBench57—

Math GPT-5.4 mini leads

GPT-5.4 mini: 45.5 (#75), Step 3: 37.6 (#148)

BenchmarkGPT-5.4 miniStep 3
LMArena Math14191366
FrontierMath (Tiers 1-3)51.2%—
FrontierMath Tier 49.8%—
OTIS Mock AIME 2024-202588.9%—
ProofBench21%—
FrontierMath (Feb 2025 set)28.3%—
FrontierMath Tier 4 (v1)2.1%—

Knowledge GPT-5.4 mini leads

GPT-5.4 mini: 51.5 (#67), Step 3: 36.8 (#164)

BenchmarkGPT-5.4 miniStep 3
LMArena Expert14351333
GPQA Diamond86.9%—
SimpleQA Verified29.4%—
Vectara Hallucination Rate5.5%—

Multimodal GPT-5.4 mini leads

GPT-5.4 mini: 39.7 (#56), Step 3: 35.5 (#86)

BenchmarkGPT-5.4 miniStep 3
LMArena Vision12451177

Multilingual GPT-5.4 mini leads

GPT-5.4 mini: 51.9 (#96), Step 3: 46.3 (#159)

BenchmarkGPT-5.4 miniStep 3
LMArena Non-English14051327
LMArena Chinese14461397
LMArena German14091371
LMArena Korean13681269
LMArena Russian14171331
LMArena Spanish14051371
LMArena French1440—
LMArena Japanese1374—

Instruction Following GPT-5.4 mini leads

GPT-5.4 mini: 74.1 (#102), Step 3: 70.4 (#164)

BenchmarkGPT-5.4 miniStep 3
LMArena Instruction Following14051332

Long Context GPT-5.4 mini leads

GPT-5.4 mini: 43.0 (#112), Step 3: 40.3 (#157)

BenchmarkGPT-5.4 miniStep 3
LMArena Longer Query14071326

Writing & Preference GPT-5.4 mini leads

GPT-5.4 mini: 64.0 (#58), Step 3: 54.3 (#151)

BenchmarkGPT-5.4 miniStep 3
LMArena Text14121350
LMArena Creative Writing13701321
LMArena Multi-Turn14291341
EQ-Bench Creative Writing1665—

Frequently asked questions

Is GPT-5.4 mini better than Step 3?

GPT-5.4 mini is the stronger model overall, scoring 45.0 to 40.5 on the Noometry Index.

Is GPT-5.4 mini or Step 3 better for coding?

GPT-5.4 mini scores higher on coding benchmarks: 45.2 versus 40.1 in the Noometry coding category.

How many benchmarks do GPT-5.4 mini and Step 3 share?

17 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and Step 3 has 17.

Related comparisons

Go deeper