Model comparison
GPT-5.4 mini vs MiniMax-M3
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 43.8 on the Noometry Index. MiniMax-M3 costs 3.2× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Last verified . 33 shared benchmarks.
MiniMaxSummary
- They share 33 benchmarks with published results for both. GPT-5.4 mini scores higher in 5 categories and MiniMax-M3 in 5 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where GPT-5.4 mini leads 29.9 to 22.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.9% for GPT-5.4 mini and 71.1% for MiniMax-M3.
- MiniMax-M3 is cheaper at $0.30 / $1.20 per million input/output tokens, against $0.75 / $4.50 for GPT-5.4 mini.
- MiniMax-M3 accepts more context: 1M tokens versus 400K.
- MiniMax-M3 has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 mini | MiniMax-M3 | |
|---|---|---|
| Provider | OpenAI | MiniMax |
| Noometry Index | 45.0 | 43.8 |
| Released | 2026-03-17 | 2026-06-01 |
| Weights | Proprietary | Open |
| Context window | 400K | 1M |
| Max output | 128K | 512K |
| Input $ / M tokens | $0.75 | $0.30 |
| Output $ / M tokens | $4.50 | $1.20 |
| Results tracked | 46 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
GPT-5.4 mini: 45.2 (#72), MiniMax-M3: 41.8 (#118)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| FrontierCode | 27% | 14.7% |
| LMArena WebDev | 1397 | 1482 |
| SciCode | 49.9% | 47.1% |
| LMArena Coding | 1438 | 1469 |
| ALE-Bench | 1,189 | 640.02 |
| WeirdML | 60.3% | — |
Agentic & Tool Use GPT-5.4 mini leads
GPT-5.4 mini: 29.9 (#81), MiniMax-M3: 22.6 (#130)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| APEX-Agents | — | 37.7% |
| OSWorld 2.0 | — | 4.6% |
| DeepResearch Bench | 36.3% | — |
| GBAEval | — | 0.9% |
| Vending-Bench 2 | — | 2,158 |
Reasoning Too close to call
GPT-5.4 mini: 30.4 (#85), MiniMax-M3: 30.1 (#87)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| NYT Connections (extended) | 61.8% | 65.1% |
| CritPt | 10% | 3.7% |
| Chess Puzzles | 24% | 14% |
| LMArena Hard Prompts | 1424 | 1447 |
| Mystery Game Puzzles | 11% | 8% |
| DTBench | 80% | 78.9% |
| LMCA | 40.8% | 33.7% |
| Epoch Capabilities Index | 148.84 | 146.95 |
| ForecastBench | 57 | 61.4 |
| ARC-AGI-2 | 18.9% | — |
| SimpleBench | — | 45.8% |
| Kagi LLM Benchmark | 37.9% | — |
| ARC-AGI-1 | 63.7% | — |
| Thematic Generalization | 61.7% | — |
| Surface Evolver Bench | — | 55% |
Math GPT-5.4 mini leads
GPT-5.4 mini: 45.5 (#75), MiniMax-M3: 40.0 (#95)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 71.1% |
| ProofBench | 21% | 18% |
| LMArena Math | 1419 | 1429 |
| FrontierMath (Tiers 1-3) | 51.2% | — |
| FrontierMath Tier 4 | 9.8% | — |
| FrontierMath (Feb 2025 set) | 28.3% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge MiniMax-M3 leads
GPT-5.4 mini: 51.5 (#67), MiniMax-M3: 58.4 (#35)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| GPQA Diamond | 86.9% | 90.9% |
| LMArena Expert | 1435 | 1461 |
| SimpleQA Verified | 29.4% | — |
| Vectara Hallucination Rate | 5.5% | — |
Multimodal Too close to call
GPT-5.4 mini: 39.7 (#56), MiniMax-M3: 40.2 (#51)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| LMArena Vision | 1245 | 1253 |
| LMArena Document | — | 1435 |
Multilingual MiniMax-M3 leads
GPT-5.4 mini: 51.9 (#96), MiniMax-M3: 53.0 (#75)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| LMArena Non-English | 1405 | 1420 |
| LMArena Chinese | 1446 | 1463 |
| LMArena French | 1440 | 1447 |
| LMArena German | 1409 | 1426 |
| LMArena Japanese | 1374 | 1381 |
| LMArena Korean | 1368 | 1372 |
| LMArena Russian | 1417 | 1428 |
| LMArena Spanish | 1405 | 1432 |
Instruction Following MiniMax-M3 leads
GPT-5.4 mini: 74.1 (#102), MiniMax-M3: 75.5 (#62)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| LMArena Instruction Following | 1405 | 1433 |
Long Context MiniMax-M3 leads
GPT-5.4 mini: 43.0 (#112), MiniMax-M3: 44.2 (#72)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| LMArena Longer Query | 1407 | 1445 |
Writing & Preference GPT-5.4 mini leads
GPT-5.4 mini: 64.0 (#58), MiniMax-M3: 62.1 (#83)
| Benchmark | GPT-5.4 mini | MiniMax-M3 |
|---|---|---|
| LMArena Text | 1412 | 1433 |
| LMArena Creative Writing | 1370 | 1404 |
| LMArena Multi-Turn | 1429 | 1442 |
| EQ-Bench Creative Writing | 1665 | — |
| EQ-Bench 4 | — | 1150 |
Frequently asked questions
Is GPT-5.4 mini better than MiniMax-M3?
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 43.8 on the Noometry Index. MiniMax-M3 costs 3.2× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 mini or MiniMax-M3?
MiniMax-M3 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; GPT-5.4 mini lists at $0.75 and $4.50.
Is GPT-5.4 mini or MiniMax-M3 better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.2 versus 41.8 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M3 does, with 1M tokens against 400K.
How many benchmarks do GPT-5.4 mini and MiniMax-M3 share?
33 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and MiniMax-M3 has 41.