Model comparison
MiniMax-M3 vs Step 5 Preview
Step 5 Preview is the stronger model overall, scoring 47.9 to 43.8 on the Noometry Index. MiniMax-M3 costs 2.7× less per token, which makes it the better buy when Step 5 Preview's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
MiniMax
StepFunSummary
- They share 18 benchmarks with published results for both. MiniMax-M3 scores higher in 1 category and Step 5 Preview in 8 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where MiniMax-M3 leads 58.4 to 41.2.
- The biggest single-benchmark swing is ProofBench: 18% for MiniMax-M3 and 42% for Step 5 Preview.
- MiniMax-M3 is cheaper at $0.30 / $1.20 per million input/output tokens, against $1 / $2.70 for Step 5 Preview.
- Step 5 Preview accepts more context: 1.02M tokens versus 1M.
- MiniMax-M3 has downloadable open weights; the other is API-only.
Side by side
| MiniMax-M3 | Step 5 Preview | |
|---|---|---|
| Provider | MiniMax | StepFun |
| Noometry Index | 43.8 | 47.9 |
| Released | 2026-06-01 | 2026-09-16 |
| Weights | Open | Proprietary |
| Context window | 1M | 1.02M |
| Max output | 512K | 66K |
| Input $ / M tokens | $0.30 | $1 |
| Output $ / M tokens | $1.20 | $2.70 |
| Results tracked | 41 | 18 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Step 5 Preview leads
MiniMax-M3: 41.8 (#118), Step 5 Preview: 51.8 (#36)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena WebDev | 1482 | 1564 |
| SciCode | 47.1% | 58.9% |
| LMArena Coding | 1469 | 1480 |
| FrontierCode | 14.7% | — |
| ALE-Bench | 640.02 | — |
Agentic & Tool Use Not comparable
MiniMax-M3: 22.6 (#130), Step 5 Preview: —
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| APEX-Agents | 37.7% | — |
| OSWorld 2.0 | 4.6% | — |
| GBAEval | 0.9% | — |
| Vending-Bench 2 | 2,158 | — |
Reasoning Step 5 Preview leads
MiniMax-M3: 30.1 (#87), Step 5 Preview: 40.0 (#57)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| CritPt | 3.7% | 20.9% |
| LMArena Hard Prompts | 1447 | 1465 |
| SimpleBench | 45.8% | — |
| NYT Connections (extended) | 65.1% | — |
| Chess Puzzles | 14% | — |
| Mystery Game Puzzles | 8% | — |
| DTBench | 78.9% | — |
| LMCA | 33.7% | — |
| Surface Evolver Bench | 55% | — |
| Epoch Capabilities Index | 146.95 | — |
| ForecastBench | 61.4 | — |
Math Step 5 Preview leads
MiniMax-M3: 40.0 (#95), Step 5 Preview: 46.1 (#72)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| ProofBench | 18% | 42% |
| LMArena Math | 1429 | 1470 |
| OTIS Mock AIME 2024-2025 | 71.1% | — |
Knowledge MiniMax-M3 leads
MiniMax-M3: 58.4 (#35), Step 5 Preview: 41.2 (#112)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena Expert | 1461 | 1470 |
| GPQA Diamond | 90.9% | — |
Multimodal Too close to call
MiniMax-M3: 40.2 (#51), Step 5 Preview: 41.0 (#41)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena Vision | 1253 | 1267 |
| LMArena Document | 1435 | — |
Multilingual Too close to call
MiniMax-M3: 53.0 (#75), Step 5 Preview: 53.8 (#54)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena Non-English | 1420 | 1432 |
| LMArena Chinese | 1463 | 1519 |
| LMArena Russian | 1428 | 1436 |
| LMArena Spanish | 1432 | 1439 |
| LMArena French | 1447 | — |
| LMArena German | 1426 | — |
| LMArena Japanese | 1381 | — |
| LMArena Korean | 1372 | — |
Instruction Following Too close to call
MiniMax-M3: 75.5 (#62), Step 5 Preview: 76.0 (#50)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena Instruction Following | 1433 | 1444 |
Long Context Too close to call
MiniMax-M3: 44.2 (#72), Step 5 Preview: 44.5 (#64)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena Longer Query | 1445 | 1455 |
Writing & Preference Too close to call
MiniMax-M3: 62.1 (#83), Step 5 Preview: 62.8 (#71)
| Benchmark | MiniMax-M3 | Step 5 Preview |
|---|---|---|
| LMArena Text | 1433 | 1442 |
| LMArena Creative Writing | 1404 | 1410 |
| LMArena Multi-Turn | 1442 | 1447 |
| EQ-Bench 4 | 1150 | — |
Frequently asked questions
Is MiniMax-M3 better than Step 5 Preview?
Step 5 Preview is the stronger model overall, scoring 47.9 to 43.8 on the Noometry Index. MiniMax-M3 costs 2.7× less per token, which makes it the better buy when Step 5 Preview's lead doesn't matter for your workload.
Which is cheaper, MiniMax-M3 or Step 5 Preview?
MiniMax-M3 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; Step 5 Preview lists at $1 and $2.70.
Is MiniMax-M3 or Step 5 Preview better for coding?
Step 5 Preview scores higher on coding benchmarks: 51.8 versus 41.8 in the Noometry coding category.
Which has the bigger context window?
Step 5 Preview does, with 1.02M tokens against 1M.
How many benchmarks do MiniMax-M3 and Step 5 Preview share?
18 benchmarks have published results for both models. MiniMax-M3 has 41 scored results on Noometry and Step 5 Preview has 18.