Model comparison
MiniMax-M2.5 vs o1
o1 is the stronger model overall, scoring 40.9 to 38.3 on the Noometry Index. MiniMax-M2.5 costs 50× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Last verified . 19 shared benchmarks.
MiniMaxSummary
- They share 19 benchmarks with published results for both. MiniMax-M2.5 scores higher in 2 categories and o1 in 7 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where o1 leads 50.3 to 37.5.
- The biggest single-benchmark swing is ARC-AGI-1: 63.7% for MiniMax-M2.5 and 30.7% for o1.
- MiniMax-M2.5 is cheaper at $0.30 / $1.20 per million input/output tokens, against $15 / $60 for o1.
- MiniMax-M2.5 accepts more context: 205K tokens versus 200K.
- MiniMax-M2.5 has downloadable open weights; the other is API-only.
Side by side
| MiniMax-M2.5 | o1 | |
|---|---|---|
| Provider | MiniMax | OpenAI |
| Noometry Index | 38.3 | 40.9 |
| Released | 2026-02-12 | 2024-09-12 |
| Weights | Open | Proprietary |
| Context window | 205K | 200K |
| Max output | 131K | 100K |
| Input $ / M tokens | $0.30 | $15 |
| Output $ / M tokens | $1.20 | $60 |
| Results tracked | 33 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M2.5 leads
MiniMax-M2.5: 48.1 (#58), o1: 46.1 (#70)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Coding | 1381 | 1367 |
| SWE-bench Verified (bash only) | 75.8% | — |
| Aider Polyglot | — | 61.7% |
| LMArena WebDev | 1387 | — |
| SWE-bench Multilingual | 68.3% | — |
| WeirdML | — | 47.6% |
| LiveBench Coding | — | 69.7% |
| CadEval | — | 56% |
| ALE-Bench | 618.17 | — |
| HumanEval+ | — | 89% |
| MBPP+ | — | 80.2% |
Agentic & Tool Use MiniMax-M2.5 leads
MiniMax-M2.5: 30.4 (#77), o1: 24.6 (#117)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| Terminal-Bench | 42.7% | — |
| Cybench | — | 10% |
| METR Time Horizons | — | 51.1% |
| Vending-Bench 2 | -23.16 | — |
Reasoning o1 leads
MiniMax-M2.5: 17.5 (#292), o1: 27.9 (#111)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| ARC-AGI-1 | 63.7% | 30.7% |
| LMArena Hard Prompts | 1372 | 1371 |
| Epoch Capabilities Index | 146.68 | 141.91 |
| ARC-AGI-2 | 4.9% | — |
| SimpleBench | — | 41.7% |
| Kagi LLM Benchmark | 55.2% | — |
| NYT Connections (extended) | 16.8% | — |
| Chess Puzzles | — | 15% |
| EnigmaEval | — | 5.7% |
| LiveBench Reasoning | — | 91.6% |
| DTBench | — | 74.7% |
| LiveBench Data Analysis | — | 65.5% |
| LMCA | — | 22.3% |
| LiveBench | — | 75.7% |
Math o1 leads
MiniMax-M2.5: 26.9 (#253), o1: 36.1 (#175)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Math | 1378 | 1388 |
| FrontierMath (Tiers 1-3) | — | 14.7% |
| OTIS Mock AIME 2024-2025 | — | 73.3% |
| ProofBench | 4% | — |
| LiveBench Math | — | 80.3% |
| MATH Level 5 | — | 94.7% |
| FrontierMath (Feb 2025 set) | — | 9.3% |
Knowledge o1 leads
MiniMax-M2.5: 39.2 (#135), o1: 41.5 (#110)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Expert | 1379 | 1361 |
| GPQA Diamond | — | 76.8% |
| Humanity's Last Exam | — | 8% |
| SimpleQA Verified | — | 41.1% |
| Confabulations | — | 11.7% |
| Vectara Hallucination Rate | 9.1% | — |
Multimodal Not comparable
MiniMax-M2.5: —, o1: 34.2 (#93)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Vision | — | 1168 |
| GeoBench | — | 80% |
| VPCT | — | 37% |
| SpatialViz-Bench | — | 41.4% |
Multilingual o1 leads
MiniMax-M2.5: 47.1 (#152), o1: 48.6 (#142)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Non-English | 1338 | 1358 |
| LMArena Chinese | 1393 | 1394 |
| LMArena French | 1362 | 1344 |
| LMArena German | 1362 | 1337 |
| LMArena Japanese | 1171 | 1346 |
| LMArena Korean | 1232 | 1396 |
| LMArena Russian | 1358 | 1356 |
| LMArena Spanish | 1354 | 1345 |
Instruction Following o1 leads
MiniMax-M2.5: 71.5 (#148), o1: 74.8 (#86)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Instruction Following | 1353 | 1367 |
| LiveBench Instruction Following | — | 81.5% |
Long Context o1 leads
MiniMax-M2.5: 37.5 (#216), o1: 50.3 (#9)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Longer Query | 1366 | 1378 |
| Fiction.LiveBench | — | 83.3% |
| CL-bench | 11.4% | — |
| CL-bench Life | 6.3% | — |
Writing & Preference o1 leads
MiniMax-M2.5: 53.9 (#153), o1: 55.6 (#144)
| Benchmark | MiniMax-M2.5 | o1 |
|---|---|---|
| LMArena Text | 1359 | 1366 |
| LMArena Creative Writing | 1331 | 1348 |
| LMArena Multi-Turn | 1364 | 1369 |
| Short-Story Creative Writing | — | 70.2% |
| EQ-Bench Creative Writing | 1361 | — |
| LiveBench Language | — | 65.4% |
Frequently asked questions
Is MiniMax-M2.5 better than o1?
o1 is the stronger model overall, scoring 40.9 to 38.3 on the Noometry Index. MiniMax-M2.5 costs 50× less per token, which makes it the better buy when o1's lead doesn't matter for your workload.
Which is cheaper, MiniMax-M2.5 or o1?
MiniMax-M2.5 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; o1 lists at $15 and $60.
Is MiniMax-M2.5 or o1 better for coding?
MiniMax-M2.5 scores higher on coding benchmarks: 48.1 versus 46.1 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M2.5 does, with 205K tokens against 200K.
How many benchmarks do MiniMax-M2.5 and o1 share?
19 benchmarks have published results for both models. MiniMax-M2.5 has 33 scored results on Noometry and o1 has 52.