Model comparison
MiniMax-M2 vs MiniMax-M3
MiniMax-M3 is the stronger model overall, scoring 43.8 to 37.4 on the Noometry Index.
Last verified . 18 shared benchmarks.
MiniMaxSummary
- They share 18 benchmarks with published results for both. MiniMax-M2 scores higher in 1 category and MiniMax-M3 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where MiniMax-M3 leads 58.4 to 37.0.
- The biggest single-benchmark swing is NYT Connections (extended): 14.8% for MiniMax-M2 and 65.1% for MiniMax-M3.
- Both cost about the same: $0.30 input and $1.20 output per million tokens.
- MiniMax-M3 accepts more context: 1M tokens versus 205K.
Side by side
| MiniMax-M2 | MiniMax-M3 | |
|---|---|---|
| Provider | MiniMax | MiniMax |
| Noometry Index | 37.4 | 43.8 |
| Released | 2025-10-27 | 2026-06-01 |
| Weights | Open | Open |
| Context window | 205K | 1M |
| Max output | 131K | 512K |
| Input $ / M tokens | $0.30 | $0.30 |
| Output $ / M tokens | $1.20 | $1.20 |
| Results tracked | 21 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M3 leads
MiniMax-M2: 39.3 (#159), MiniMax-M3: 41.8 (#118)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena WebDev | 1297 | 1482 |
| LMArena Coding | 1370 | 1469 |
| FrontierCode | — | 14.7% |
| SWE-bench Verified (bash only) | 61% | — |
| SciCode | — | 47.1% |
| ALE-Bench | — | 640.02 |
Agentic & Tool Use MiniMax-M2 leads
MiniMax-M2: 25.1 (#109), MiniMax-M3: 22.6 (#130)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| Vending-Bench 2 | 160.6 | 2,158 |
| Terminal-Bench | 30% | — |
| APEX-Agents | — | 37.7% |
| OSWorld 2.0 | — | 4.6% |
| GBAEval | — | 0.9% |
Reasoning MiniMax-M3 leads
MiniMax-M2: 19.4 (#258), MiniMax-M3: 30.1 (#87)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| NYT Connections (extended) | 14.8% | 65.1% |
| LMArena Hard Prompts | 1357 | 1447 |
| SimpleBench | — | 45.8% |
| Kagi LLM Benchmark | 57.8% | — |
| CritPt | — | 3.7% |
| 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 MiniMax-M3 leads
MiniMax-M2: 37.3 (#160), MiniMax-M3: 40.0 (#95)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Math | 1352 | 1429 |
| OTIS Mock AIME 2024-2025 | — | 71.1% |
| ProofBench | — | 18% |
Knowledge MiniMax-M3 leads
MiniMax-M2: 37.0 (#163), MiniMax-M3: 58.4 (#35)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Expert | 1337 | 1461 |
| GPQA Diamond | — | 90.9% |
Multimodal Not comparable
MiniMax-M2: —, MiniMax-M3: 40.2 (#51)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Vision | — | 1253 |
| LMArena Document | — | 1435 |
Multilingual MiniMax-M3 leads
MiniMax-M2: 45.3 (#171), MiniMax-M3: 53.0 (#75)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Non-English | 1313 | 1420 |
| LMArena Chinese | 1366 | 1463 |
| LMArena French | 1335 | 1447 |
| LMArena German | 1355 | 1426 |
| LMArena Russian | 1331 | 1428 |
| LMArena Spanish | 1326 | 1432 |
| LMArena Japanese | — | 1381 |
| LMArena Korean | — | 1372 |
Instruction Following MiniMax-M3 leads
MiniMax-M2: 70.2 (#166), MiniMax-M3: 75.5 (#62)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Instruction Following | 1328 | 1433 |
Long Context MiniMax-M3 leads
MiniMax-M2: 40.5 (#153), MiniMax-M3: 44.2 (#72)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Longer Query | 1331 | 1445 |
Writing & Preference MiniMax-M3 leads
MiniMax-M2: 53.0 (#162), MiniMax-M3: 62.1 (#83)
| Benchmark | MiniMax-M2 | MiniMax-M3 |
|---|---|---|
| LMArena Text | 1340 | 1433 |
| LMArena Creative Writing | 1286 | 1404 |
| LMArena Multi-Turn | 1361 | 1442 |
| EQ-Bench 4 | — | 1150 |
Frequently asked questions
Is MiniMax-M2 better than MiniMax-M3?
MiniMax-M3 is the stronger model overall, scoring 43.8 to 37.4 on the Noometry Index.
Which is cheaper, MiniMax-M2 or MiniMax-M3?
MiniMax-M3 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; MiniMax-M2 lists at $0.30 and $1.20.
Is MiniMax-M2 or MiniMax-M3 better for coding?
MiniMax-M3 scores higher on coding benchmarks: 41.8 versus 39.3 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M3 does, with 1M tokens against 205K.
How many benchmarks do MiniMax-M2 and MiniMax-M3 share?
18 benchmarks have published results for both models. MiniMax-M2 has 21 scored results on Noometry and MiniMax-M3 has 41.