Model comparison
GPT-4 vs Step 3
Step 3 is the stronger model overall, scoring 40.5 to 29.1 on the Noometry Index.
Last verified . 15 shared benchmarks.
StepFunSummary
- They share 15 benchmarks with published results for both. GPT-4 scores higher in 0 categories and Step 3 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Step 3 leads 37.6 to 10.8.
- Step 3 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 | Step 3 | |
|---|---|---|
| Provider | OpenAI | StepFun |
| Noometry Index | 29.1 | 40.5 |
| Released | 2023-03-14 | — |
| Weights | Proprietary | Open |
| Context window | 8K | — |
| Max output | 8K | — |
| Input $ / M tokens | $30 | — |
| Output $ / M tokens | $60 | — |
| Results tracked | 38 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 3 leads
GPT-4: 31.6 (#283), Step 3: 40.1 (#147)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Coding | 1254 | 1367 |
| WeirdML | 12.4% | — |
| BigCodeBench Instruct | 46% | — |
| BigCodeBench Complete | 57.2% | — |
| HumanEval+ | 79.3% | — |
Agentic & Tool Use Not comparable
GPT-4: —, Step 3: —
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| METR Time Horizons | 36.1% | — |
Reasoning Step 3 leads
GPT-4: 17.8 (#289), Step 3: 28.4 (#105)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Hard Prompts | 1241 | 1355 |
| Kagi LLM Benchmark | — | 62.3% |
| Chess Puzzles | 4% | — |
| Mystery Game Puzzles | 12% | — |
| DTBench | 62.7% | — |
| LMCA | 17.1% | — |
| BIG-Bench Hard | 75.1% | — |
| Epoch Capabilities Index | 125.89 | — |
| ForecastBench | 57.8 | — |
| HellaSwag | 95.3% | — |
| WinoGrande | 87.5% | — |
Math Step 3 leads
GPT-4: 10.8 (#309), Step 3: 37.6 (#148)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Math | 1269 | 1366 |
| OTIS Mock AIME 2024-2025 | 1.1% | — |
| MATH Level 5 | 23% | — |
| GSM8K | 92% | — |
Knowledge Step 3 leads
GPT-4: 18.4 (#282), Step 3: 36.8 (#164)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Expert | 1211 | 1333 |
| GPQA Diamond | 35.7% | — |
| MMLU | 86.4% | — |
| TriviaQA | 84.8% | — |
Multimodal Not comparable
GPT-4: —, Step 3: 35.5 (#86)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Vision | — | 1177 |
Multilingual Step 3 leads
GPT-4: 40.6 (#215), Step 3: 46.3 (#159)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Non-English | 1246 | 1327 |
| LMArena Chinese | 1242 | 1397 |
| LMArena German | 1251 | 1371 |
| LMArena Korean | 1184 | 1269 |
| LMArena Russian | 1251 | 1331 |
| LMArena Spanish | 1261 | 1371 |
| LMArena French | 1283 | — |
| LMArena Japanese | 1209 | — |
Instruction Following Step 3 leads
GPT-4: 65.3 (#222), Step 3: 70.4 (#164)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1241 | 1332 |
Long Context Step 3 leads
GPT-4: 37.7 (#212), Step 3: 40.3 (#157)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Longer Query | 1244 | 1326 |
Writing & Preference Step 3 leads
GPT-4: 34.9 (#268), Step 3: 54.3 (#151)
| Benchmark | GPT-4 | Step 3 |
|---|---|---|
| LMArena Text | 1263 | 1350 |
| LMArena Creative Writing | 1244 | 1321 |
| LMArena Multi-Turn | 1257 | 1341 |
| EQ-Bench Creative Writing | 752 | — |
Frequently asked questions
Is GPT-4 better than Step 3?
Step 3 is the stronger model overall, scoring 40.5 to 29.1 on the Noometry Index.
Is GPT-4 or Step 3 better for coding?
Step 3 scores higher on coding benchmarks: 40.1 versus 31.6 in the Noometry coding category.
How many benchmarks do GPT-4 and Step 3 share?
15 benchmarks have published results for both models. GPT-4 has 38 scored results on Noometry and Step 3 has 17.