Model comparison
Gemma 3 4B vs MiniMax-M2.5
MiniMax-M2.5 is the stronger model overall, scoring 38.3 to 28.1 on the Noometry Index. Gemma 3 4B costs 10× less per token, which makes it the better buy when MiniMax-M2.5's lead doesn't matter for your workload.
Last verified . 16 shared benchmarks.
Google
MiniMaxSummary
- They share 16 benchmarks with published results for both. Gemma 3 4B scores higher in 1 category and MiniMax-M2.5 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where MiniMax-M2.5 leads 39.2 to 11.8.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 25.2% for Gemma 3 4B and 55.2% for MiniMax-M2.5.
- Gemma 3 4B is cheaper at $0.04 / $0.08 per million input/output tokens, against $0.30 / $1.20 for MiniMax-M2.5.
- MiniMax-M2.5 accepts more context: 205K tokens versus 131K.
Side by side
| Gemma 3 4B | MiniMax-M2.5 | |
|---|---|---|
| Provider | MiniMax | |
| Noometry Index | 28.1 | 38.3 |
| Released | 2025-03-12 | 2026-02-12 |
| Weights | Open | Open |
| Context window | 131K | 205K |
| Max output | 4K | 131K |
| Input $ / M tokens | $0.04 | $0.30 |
| Output $ / M tokens | $0.08 | $1.20 |
| Results tracked | 22 | 33 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M2.5 leads
Gemma 3 4B: 35.9 (#215), MiniMax-M2.5: 48.1 (#58)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| LMArena Coding | 1230 | 1381 |
| SWE-bench Verified (bash only) | — | 75.8% |
| LMArena WebDev | — | 1387 |
| SWE-bench Multilingual | — | 68.3% |
| ALE-Bench | — | 618.17 |
Agentic & Tool Use MiniMax-M2.5 leads
Gemma 3 4B: 20.9 (#142), MiniMax-M2.5: 30.4 (#77)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| Terminal-Bench | — | 42.7% |
| Berkeley Function Calling Leaderboard | 19.6% | — |
| Vending-Bench 2 | — | -23.16 |
Reasoning MiniMax-M2.5 leads
Gemma 3 4B: 13.2 (#335), MiniMax-M2.5: 17.5 (#292)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| Kagi LLM Benchmark | 25.2% | 55.2% |
| LMArena Hard Prompts | 1253 | 1372 |
| Epoch Capabilities Index | 116.02 | 146.68 |
| ARC-AGI-2 | — | 4.9% |
| NYT Connections (extended) | — | 16.8% |
| ARC-AGI-1 | — | 63.7% |
| Chess Puzzles | 0% | — |
| DTBench | 50.9% | — |
| LMCA | 2.8% | — |
Math MiniMax-M2.5 leads
Gemma 3 4B: 16.8 (#292), MiniMax-M2.5: 26.9 (#253)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| LMArena Math | 1239 | 1378 |
| OTIS Mock AIME 2024-2025 | 7.5% | — |
| ProofBench | — | 4% |
Knowledge MiniMax-M2.5 leads
Gemma 3 4B: 11.8 (#299), MiniMax-M2.5: 39.2 (#135)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| Vectara Hallucination Rate | 6.4% | 9.1% |
| LMArena Expert | 1223 | 1379 |
| GPQA Diamond | 23.2% | — |
Multilingual MiniMax-M2.5 leads
Gemma 3 4B: 42.5 (#194), MiniMax-M2.5: 47.1 (#152)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| LMArena Non-English | 1273 | 1338 |
| LMArena German | 1281 | 1362 |
| LMArena Russian | 1294 | 1358 |
| LMArena Chinese | — | 1393 |
| LMArena French | — | 1362 |
| LMArena Japanese | — | 1171 |
| LMArena Korean | — | 1232 |
| LMArena Spanish | — | 1354 |
Instruction Following MiniMax-M2.5 leads
Gemma 3 4B: 65.2 (#225), MiniMax-M2.5: 71.5 (#148)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| LMArena Instruction Following | 1239 | 1353 |
Long Context Gemma 3 4B leads
Gemma 3 4B: 38.7 (#194), MiniMax-M2.5: 37.5 (#216)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| LMArena Longer Query | 1273 | 1366 |
| CL-bench | — | 11.4% |
| CL-bench Life | — | 6.3% |
Writing & Preference MiniMax-M2.5 leads
Gemma 3 4B: 42.0 (#239), MiniMax-M2.5: 53.9 (#153)
| Benchmark | Gemma 3 4B | MiniMax-M2.5 |
|---|---|---|
| LMArena Text | 1291 | 1359 |
| LMArena Creative Writing | 1271 | 1331 |
| EQ-Bench Creative Writing | 1068 | 1361 |
| LMArena Multi-Turn | 1255 | 1364 |
Frequently asked questions
Is Gemma 3 4B better than MiniMax-M2.5?
MiniMax-M2.5 is the stronger model overall, scoring 38.3 to 28.1 on the Noometry Index. Gemma 3 4B costs 10× less per token, which makes it the better buy when MiniMax-M2.5's lead doesn't matter for your workload.
Which is cheaper, Gemma 3 4B or MiniMax-M2.5?
Gemma 3 4B is cheaper. It lists at $0.04 per million input tokens and $0.08 per million output tokens; MiniMax-M2.5 lists at $0.30 and $1.20.
Is Gemma 3 4B or MiniMax-M2.5 better for coding?
MiniMax-M2.5 scores higher on coding benchmarks: 48.1 versus 35.9 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M2.5 does, with 205K tokens against 131K.
How many benchmarks do Gemma 3 4B and MiniMax-M2.5 share?
16 benchmarks have published results for both models. Gemma 3 4B has 22 scored results on Noometry and MiniMax-M2.5 has 33.