Model comparison
GPT-4 Turbo vs Step 3
Step 3 is the stronger model overall, scoring 40.5 to 30.5 on the Noometry Index.
Last verified . 16 shared benchmarks.
StepFunSummary
- They share 16 benchmarks with published results for both. GPT-4 Turbo scores higher in 0 categories and Step 3 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Step 3 leads 37.6 to 9.0.
- Step 3 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 Turbo | Step 3 | |
|---|---|---|
| Provider | OpenAI | StepFun |
| Noometry Index | 30.5 | 40.5 |
| Released | 2023-11-06 | — |
| Weights | Proprietary | Open |
| Context window | 128K | — |
| Max output | 4K | — |
| Input $ / M tokens | $10 | — |
| Output $ / M tokens | $30 | — |
| Results tracked | 36 | 17 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 3 leads
GPT-4 Turbo: 33.8 (#249), Step 3: 40.1 (#147)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Coding | 1268 | 1367 |
| WeirdML | 18% | — |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 58.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73.3% | — |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, Step 3: —
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| METR Time Horizons | 36.7% | — |
Reasoning Step 3 leads
GPT-4 Turbo: 15.3 (#317), Step 3: 28.4 (#105)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Hard Prompts | 1251 | 1355 |
| SimpleBench | 25.1% | — |
| Kagi LLM Benchmark | — | 62.3% |
| Chess Puzzles | 6% | — |
| DTBench | 61.6% | — |
| LMCA | 9.8% | — |
| Epoch Capabilities Index | 127.25 | — |
| ForecastBench | 59.4 | — |
Math Step 3 leads
GPT-4 Turbo: 9.0 (#322), Step 3: 37.6 (#148)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Math | 1272 | 1366 |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| OTIS Mock AIME 2024-2025 | 6.7% | — |
| MATH Level 5 | 46.7% | — |
Knowledge Step 3 leads
GPT-4 Turbo: 24.3 (#268), Step 3: 36.8 (#164)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Expert | 1223 | 1333 |
| GPQA Diamond | 46.6% | — |
| Confabulations | 28.4% | — |
| MMLU | 81.3% | — |
Multimodal Step 3 leads
GPT-4 Turbo: 30.6 (#110), Step 3: 35.5 (#86)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Vision | 1090 | 1177 |
Multilingual Step 3 leads
GPT-4 Turbo: 40.5 (#216), Step 3: 46.3 (#159)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Non-English | 1245 | 1327 |
| LMArena Chinese | 1242 | 1397 |
| LMArena German | 1259 | 1371 |
| LMArena Korean | 1187 | 1269 |
| LMArena Russian | 1259 | 1331 |
| LMArena Spanish | 1260 | 1371 |
| LMArena French | 1276 | — |
| LMArena Japanese | 1194 | — |
Instruction Following Step 3 leads
GPT-4 Turbo: 65.8 (#216), Step 3: 70.4 (#164)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Instruction Following | 1249 | 1332 |
Long Context Step 3 leads
GPT-4 Turbo: 38.0 (#206), Step 3: 40.3 (#157)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Longer Query | 1254 | 1326 |
Writing & Preference Step 3 leads
GPT-4 Turbo: 47.7 (#206), Step 3: 54.3 (#151)
| Benchmark | GPT-4 Turbo | Step 3 |
|---|---|---|
| LMArena Text | 1272 | 1350 |
| LMArena Creative Writing | 1269 | 1321 |
| LMArena Multi-Turn | 1267 | 1341 |
Frequently asked questions
Is GPT-4 Turbo better than Step 3?
Step 3 is the stronger model overall, scoring 40.5 to 30.5 on the Noometry Index.
Is GPT-4 Turbo or Step 3 better for coding?
Step 3 scores higher on coding benchmarks: 40.1 versus 33.8 in the Noometry coding category.
How many benchmarks do GPT-4 Turbo and Step 3 share?
16 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and Step 3 has 17.