Model comparison
GPT-5.4 mini vs Step 5 Preview
Step 5 Preview is the stronger model overall, scoring 47.9 to 45.0 on the Noometry Index.
Last verified . 18 shared benchmarks.
StepFunSummary
- They share 18 benchmarks with published results for both. GPT-5.4 mini scores higher in 2 categories and Step 5 Preview in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where GPT-5.4 mini leads 51.5 to 41.2.
- The biggest single-benchmark swing is ProofBench: 21% for GPT-5.4 mini and 42% for Step 5 Preview.
- Step 5 Preview is cheaper at $1 / $2.70 per million input/output tokens, against $0.75 / $4.50 for GPT-5.4 mini.
- Step 5 Preview accepts more context: 1.02M tokens versus 400K.
Side by side
| GPT-5.4 mini | Step 5 Preview | |
|---|---|---|
| Provider | OpenAI | StepFun |
| Noometry Index | 45.0 | 47.9 |
| Released | 2026-03-17 | 2026-09-16 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 1.02M |
| Max output | 128K | 66K |
| Input $ / M tokens | $0.75 | $1 |
| Output $ / M tokens | $4.50 | $2.70 |
| Results tracked | 46 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
GPT-5.4 mini: 45.2 (#72), Step 5 Preview: 51.8 (#36)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena WebDev | 1397 | 1564 |
| SciCode | 49.9% | 58.9% |
| LMArena Coding | 1438 | 1480 |
| FrontierCode | 27% | — |
| WeirdML | 60.3% | — |
| ALE-Bench | 1,189 | — |
Agentic & Tool Use Not comparable
GPT-5.4 mini: 29.9 (#81), Step 5 Preview: —
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| DeepResearch Bench | 36.3% | — |
Reasoning Step 5 Preview leads
GPT-5.4 mini: 30.4 (#85), Step 5 Preview: 40.0 (#57)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| CritPt | 10% | 20.9% |
| LMArena Hard Prompts | 1424 | 1465 |
| ARC-AGI-2 | 18.9% | — |
| Kagi LLM Benchmark | 37.9% | — |
| NYT Connections (extended) | 61.8% | — |
| ARC-AGI-1 | 63.7% | — |
| Chess Puzzles | 24% | — |
| Thematic Generalization | 61.7% | — |
| Mystery Game Puzzles | 11% | — |
| DTBench | 80% | — |
| LMCA | 40.8% | — |
| Epoch Capabilities Index | 148.84 | — |
| ForecastBench | 57 | — |
Math Too close to call
GPT-5.4 mini: 45.5 (#75), Step 5 Preview: 46.1 (#72)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| ProofBench | 21% | 42% |
| LMArena Math | 1419 | 1470 |
| FrontierMath (Tiers 1-3) | 51.2% | — |
| FrontierMath Tier 4 | 9.8% | — |
| OTIS Mock AIME 2024-2025 | 88.9% | — |
| 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 5 Preview: 41.2 (#112)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1435 | 1470 |
| GPQA Diamond | 86.9% | — |
| SimpleQA Verified | 29.4% | — |
| Vectara Hallucination Rate | 5.5% | — |
Multimodal Step 5 Preview leads
GPT-5.4 mini: 39.7 (#56), Step 5 Preview: 41.0 (#41)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1245 | 1267 |
Multilingual Step 5 Preview leads
GPT-5.4 mini: 51.9 (#96), Step 5 Preview: 53.8 (#54)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1405 | 1432 |
| LMArena Chinese | 1446 | 1519 |
| LMArena Russian | 1417 | 1436 |
| LMArena Spanish | 1405 | 1439 |
| LMArena French | 1440 | — |
| LMArena German | 1409 | — |
| LMArena Japanese | 1374 | — |
| LMArena Korean | 1368 | — |
Instruction Following Step 5 Preview leads
GPT-5.4 mini: 74.1 (#102), Step 5 Preview: 76.0 (#50)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1405 | 1444 |
Long Context Step 5 Preview leads
GPT-5.4 mini: 43.0 (#112), Step 5 Preview: 44.5 (#64)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1407 | 1455 |
Writing & Preference GPT-5.4 mini leads
GPT-5.4 mini: 64.0 (#58), Step 5 Preview: 62.8 (#71)
| Benchmark | GPT-5.4 mini | Step 5 Preview |
|---|---|---|
| LMArena Text | 1412 | 1442 |
| LMArena Creative Writing | 1370 | 1410 |
| LMArena Multi-Turn | 1429 | 1447 |
| EQ-Bench Creative Writing | 1665 | — |
Frequently asked questions
Is GPT-5.4 mini better than Step 5 Preview?
Step 5 Preview is the stronger model overall, scoring 47.9 to 45.0 on the Noometry Index.
Which is cheaper, GPT-5.4 mini or Step 5 Preview?
Step 5 Preview is cheaper. It lists at $1 per million input tokens and $2.70 per million output tokens; GPT-5.4 mini lists at $0.75 and $4.50.
Is GPT-5.4 mini or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 45.2 in the Noometry coding category.
Which has the bigger context window?
Step 5 Preview does, with 1.02M tokens against 400K.
How many benchmarks do GPT-5.4 mini and Step 5 Preview share?
18 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and Step 5 Preview has 18.