Model comparison
MiniMax-M2.1 vs Phi-4
MiniMax-M2.1 is the stronger model overall, scoring 38.4 to 30.9 on the Noometry Index. Phi-4 costs 6.0× less per token, which makes it the better buy when MiniMax-M2.1's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
MiniMax
MicrosoftSummary
- They share 18 benchmarks with published results for both. MiniMax-M2.1 scores higher in 8 categories and Phi-4 in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where MiniMax-M2.1 leads 58.5 to 40.7.
- The biggest single-benchmark swing is Vectara Hallucination Rate: 11.8% for MiniMax-M2.1 and 3.7% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $0.30 / $1.20 for MiniMax-M2.1.
- MiniMax-M2.1 accepts more context: 205K tokens versus 128K.
Side by side
| MiniMax-M2.1 | Phi-4 | |
|---|---|---|
| Provider | MiniMax | Microsoft |
| Noometry Index | 38.4 | 30.9 |
| Released | 2025-12-23 | 2024-12-11 |
| Weights | Open | Open |
| Context window | 205K | 128K |
| Max output | 131K | 4K |
| Input $ / M tokens | $0.30 | $0.07 |
| Output $ / M tokens | $1.20 | $0.14 |
| Results tracked | 22 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M2.1 leads
MiniMax-M2.1: 40.1 (#132), Phi-4: 33.8 (#237)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Coding | 1421 | 1231 |
| LMArena WebDev | 1384 | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 623.83 | — |
Agentic & Tool Use MiniMax-M2.1 leads
MiniMax-M2.1: 27.7 (#91), Phi-4: 22.6 (#120)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| Terminal-Bench | 36.6% | — |
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
Reasoning Phi-4 leads
MiniMax-M2.1: 15.8 (#301), Phi-4: 17.0 (#287)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Hard Prompts | 1410 | 1220 |
| NYT Connections (extended) | 11.2% | — |
| Chess Puzzles | — | 1% |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| Epoch Capabilities Index | — | 130.43 |
| LiveBench | — | 41.6% |
Math MiniMax-M2.1 leads
MiniMax-M2.1: 35.6 (#142), Phi-4: 19.6 (#279)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Math | 1396 | 1246 |
| OTIS Mock AIME 2024-2025 | — | 13.8% |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
Knowledge MiniMax-M2.1 leads
MiniMax-M2.1: 37.7 (#143), Phi-4: 32.5 (#204)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| Vectara Hallucination Rate | 11.8% | 3.7% |
| LMArena Expert | 1432 | 1203 |
| GPQA Diamond | — | 56.1% |
| Confabulations | — | 29.4% |
| MMLU | — | 84.8% |
Multilingual MiniMax-M2.1 leads
MiniMax-M2.1: 50.2 (#123), Phi-4: 37.4 (#232)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Non-English | 1378 | 1197 |
| LMArena Chinese | 1430 | 1211 |
| LMArena French | 1405 | 1223 |
| LMArena German | 1380 | 1222 |
| LMArena Japanese | 1287 | 1158 |
| LMArena Korean | 1297 | 1151 |
| LMArena Russian | 1386 | 1209 |
| LMArena Spanish | 1396 | 1234 |
Instruction Following MiniMax-M2.1 leads
MiniMax-M2.1: 74.0 (#108), Phi-4: 60.6 (#246)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1399 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context MiniMax-M2.1 leads
MiniMax-M2.1: 43.2 (#98), Phi-4: 36.9 (#221)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1416 | 1217 |
Writing & Preference MiniMax-M2.1 leads
MiniMax-M2.1: 58.5 (#114), Phi-4: 40.7 (#239)
| Benchmark | MiniMax-M2.1 | Phi-4 |
|---|---|---|
| LMArena Text | 1392 | 1217 |
| LMArena Creative Writing | 1362 | 1182 |
| LMArena Multi-Turn | 1396 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is MiniMax-M2.1 better than Phi-4?
MiniMax-M2.1 is the stronger model overall, scoring 38.4 to 30.9 on the Noometry Index. Phi-4 costs 6.0× less per token, which makes it the better buy when MiniMax-M2.1's lead doesn't matter for your workload.
Which is cheaper, MiniMax-M2.1 or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; MiniMax-M2.1 lists at $0.30 and $1.20.
Is MiniMax-M2.1 or Phi-4 better for coding?
MiniMax-M2.1 scores higher on coding benchmarks: 40.1 versus 33.8 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M2.1 does, with 205K tokens against 128K.
How many benchmarks do MiniMax-M2.1 and Phi-4 share?
18 benchmarks have published results for both models. MiniMax-M2.1 has 22 scored results on Noometry and Phi-4 has 37.