Model comparison
GPT-6 Sol vs Step 5 Preview
GPT-6 Sol is the stronger model overall, scoring 61.8 to 47.9 on the Noometry Index. Step 5 Preview costs 2.8× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
StepFunSummary
- They share 18 benchmarks with published results for both. GPT-6 Sol scores higher in 6 categories and Step 5 Preview in 3 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GPT-6 Sol leads 87.2 to 46.1.
- The biggest single-benchmark swing is ProofBench: 83% for GPT-6 Sol and 42% for Step 5 Preview.
- Step 5 Preview is cheaper at $1 / $2.70 per million input/output tokens, against $2 / $10 for GPT-6 Sol.
- GPT-6 Sol accepts more context: 1.05M tokens versus 1.02M.
Side by side
| GPT-6 Sol | Step 5 Preview | |
|---|---|---|
| Provider | OpenAI | StepFun |
| Noometry Index | 61.8 | 47.9 |
| Released | 2026-09-22 | 2026-09-16 |
| Weights | Proprietary | Proprietary |
| Context window | 1.05M | 1.02M |
| Max output | 128K | 66K |
| Input $ / M tokens | $2 | $1 |
| Output $ / M tokens | $10 | $2.70 |
| Results tracked | 45 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-6 Sol leads
GPT-6 Sol: 60.1 (#11), Step 5 Preview: 51.8 (#36)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena WebDev | 1688 | 1564 |
| SciCode | 57.6% | 58.9% |
| LMArena Coding | 1447 | 1480 |
| DeepSWE | 68.8% | — |
| FrontierCode | 49.3% | — |
| ALE-Bench | 2,462 | — |
Agentic & Tool Use Not comparable
GPT-6 Sol: 37.2 (#36), Step 5 Preview: —
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| APEX-Agents | 54.3% | — |
| GDP.pdf | 26.4% | — |
| Vending-Bench 2 | 14,428 | — |
Reasoning GPT-6 Sol leads
GPT-6 Sol: 74.0 (#9), Step 5 Preview: 40.0 (#57)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| CritPt | 30.9% | 20.9% |
| LMArena Hard Prompts | 1418 | 1465 |
| ARC-AGI-2 | 89.6% | — |
| NYT Connections (extended) | 90.1% | — |
| ARC-AGI-1 | 95.5% | — |
| EBR-Bench | 53.3% | — |
| Mystery Game Puzzles | 56% | — |
| DTBench | 97.3% | — |
| LMCA | 59.1% | — |
| Epoch Capabilities Index | 162.72 | — |
Math GPT-6 Sol leads
GPT-6 Sol: 87.2 (#7), Step 5 Preview: 46.1 (#72)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| ProofBench | 83% | 42% |
| LMArena Math | 1402 | 1470 |
| FrontierMath (Tiers 1-3) | 89.8% | — |
| FrontierMath Tier 4 | 90% | — |
| OTIS Mock AIME 2024-2025 | 100% | — |
Knowledge GPT-6 Sol leads
GPT-6 Sol: 64.8 (#15), Step 5 Preview: 41.2 (#112)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1439 | 1470 |
| GPQA Diamond | 94.3% | — |
| SimpleQA Verified | 60.7% | — |
| Vectara Hallucination Rate | 6.5% | — |
Multimodal GPT-6 Sol leads
GPT-6 Sol: 47.6 (#10), Step 5 Preview: 41.0 (#41)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1245 | 1267 |
| Blueprint-Bench 2 | 36.9% | — |
| Furniture Assembly | 58.3% | — |
Multilingual Step 5 Preview leads
GPT-6 Sol: 50.5 (#118), Step 5 Preview: 53.8 (#54)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1385 | 1432 |
| LMArena Chinese | 1405 | 1519 |
| LMArena Russian | 1401 | 1436 |
| LMArena Spanish | 1384 | 1439 |
| LMArena French | 1410 | — |
| LMArena German | 1390 | — |
| LMArena Japanese | 1385 | — |
| LMArena Korean | 1341 | — |
Instruction Following Step 5 Preview leads
GPT-6 Sol: 74.5 (#94), Step 5 Preview: 76.0 (#50)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1412 | 1444 |
Long Context Step 5 Preview leads
GPT-6 Sol: 43.1 (#108), Step 5 Preview: 44.5 (#64)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1411 | 1455 |
Writing & Preference GPT-6 Sol leads
GPT-6 Sol: 71.9 (#18), Step 5 Preview: 62.8 (#71)
| Benchmark | GPT-6 Sol | Step 5 Preview |
|---|---|---|
| LMArena Text | 1395 | 1442 |
| LMArena Creative Writing | 1378 | 1410 |
| LMArena Multi-Turn | 1412 | 1447 |
| EQ-Bench Creative Writing | 2125 | — |
Frequently asked questions
Is GPT-6 Sol better than Step 5 Preview?
GPT-6 Sol is the stronger model overall, scoring 61.8 to 47.9 on the Noometry Index. Step 5 Preview costs 2.8× less per token, which makes it the better buy when GPT-6 Sol's lead doesn't matter for your workload.
Which is cheaper, GPT-6 Sol 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-6 Sol lists at $2 and $10.
Is GPT-6 Sol or Step 5 Preview better for coding?
GPT-6 Sol scores higher on coding benchmarks: 60.1 versus 51.8 in the Noometry coding category.
Which has the bigger context window?
GPT-6 Sol does, with 1.05M tokens against 1.02M.
How many benchmarks do GPT-6 Sol and Step 5 Preview share?
18 benchmarks have published results for both models. GPT-6 Sol has 45 scored results on Noometry and Step 5 Preview has 18.