Model comparison
GPT-4 Turbo vs MiniMax-M2.5
MiniMax-M2.5 is the stronger model overall, scoring 38.3 to 30.5 on the Noometry Index.
Last verified . 18 shared benchmarks.
Summary
- They share 18 benchmarks with published results for both. GPT-4 Turbo scores higher in 1 category and MiniMax-M2.5 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where MiniMax-M2.5 leads 26.9 to 9.0.
- MiniMax-M2.5 is cheaper at $0.30 / $1.20 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- MiniMax-M2.5 accepts more context: 205K tokens versus 128K.
- MiniMax-M2.5 has downloadable open weights; the other is API-only.
Side by side
| GPT-4 Turbo | MiniMax-M2.5 | |
|---|---|---|
| Provider | OpenAI | MiniMax |
| Noometry Index | 30.5 | 38.3 |
| Released | 2023-11-06 | 2026-02-12 |
| Weights | Proprietary | Open |
| Context window | 128K | 205K |
| Max output | 4K | 131K |
| Input $ / M tokens | $10 | $0.30 |
| Output $ / M tokens | $30 | $1.20 |
| Results tracked | 36 | 33 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M2.5 leads
GPT-4 Turbo: 33.8 (#249), MiniMax-M2.5: 48.1 (#58)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Coding | 1268 | 1381 |
| SWE-bench Verified (bash only) | — | 75.8% |
| LMArena WebDev | — | 1387 |
| SWE-bench Multilingual | — | 68.3% |
| WeirdML | 18% | — |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 58.2% | — |
| ALE-Bench | — | 618.17 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73.3% | — |
Agentic & Tool Use Not comparable
GPT-4 Turbo: —, MiniMax-M2.5: 30.4 (#77)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| Terminal-Bench | — | 42.7% |
| METR Time Horizons | 36.7% | — |
| Vending-Bench 2 | — | -23.16 |
Reasoning MiniMax-M2.5 leads
GPT-4 Turbo: 15.3 (#317), MiniMax-M2.5: 17.5 (#292)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Hard Prompts | 1251 | 1372 |
| Epoch Capabilities Index | 127.25 | 146.68 |
| ARC-AGI-2 | — | 4.9% |
| SimpleBench | 25.1% | — |
| Kagi LLM Benchmark | — | 55.2% |
| NYT Connections (extended) | — | 16.8% |
| ARC-AGI-1 | — | 63.7% |
| Chess Puzzles | 6% | — |
| DTBench | 61.6% | — |
| LMCA | 9.8% | — |
| ForecastBench | 59.4 | — |
Math MiniMax-M2.5 leads
GPT-4 Turbo: 9.0 (#322), MiniMax-M2.5: 26.9 (#253)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Math | 1272 | 1378 |
| FrontierMath (Tiers 1-3) | 0.7% | — |
| OTIS Mock AIME 2024-2025 | 6.7% | — |
| ProofBench | — | 4% |
| MATH Level 5 | 46.7% | — |
Knowledge MiniMax-M2.5 leads
GPT-4 Turbo: 24.3 (#268), MiniMax-M2.5: 39.2 (#135)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Expert | 1223 | 1379 |
| GPQA Diamond | 46.6% | — |
| Confabulations | 28.4% | — |
| Vectara Hallucination Rate | — | 9.1% |
| MMLU | 81.3% | — |
Multimodal Not comparable
GPT-4 Turbo: 30.6 (#110), MiniMax-M2.5: —
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Vision | 1090 | — |
Multilingual MiniMax-M2.5 leads
GPT-4 Turbo: 40.5 (#216), MiniMax-M2.5: 47.1 (#152)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Non-English | 1245 | 1338 |
| LMArena Chinese | 1242 | 1393 |
| LMArena French | 1276 | 1362 |
| LMArena German | 1259 | 1362 |
| LMArena Japanese | 1194 | 1171 |
| LMArena Korean | 1187 | 1232 |
| LMArena Russian | 1259 | 1358 |
| LMArena Spanish | 1260 | 1354 |
Instruction Following MiniMax-M2.5 leads
GPT-4 Turbo: 65.8 (#216), MiniMax-M2.5: 71.5 (#148)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Instruction Following | 1249 | 1353 |
Long Context Too close to call
GPT-4 Turbo: 38.0 (#206), MiniMax-M2.5: 37.5 (#216)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Longer Query | 1254 | 1366 |
| CL-bench | — | 11.4% |
| CL-bench Life | — | 6.3% |
Writing & Preference MiniMax-M2.5 leads
GPT-4 Turbo: 47.7 (#206), MiniMax-M2.5: 53.9 (#153)
| Benchmark | GPT-4 Turbo | MiniMax-M2.5 |
|---|---|---|
| LMArena Text | 1272 | 1359 |
| LMArena Creative Writing | 1269 | 1331 |
| LMArena Multi-Turn | 1267 | 1364 |
| EQ-Bench Creative Writing | — | 1361 |
Frequently asked questions
Is GPT-4 Turbo better than MiniMax-M2.5?
MiniMax-M2.5 is the stronger model overall, scoring 38.3 to 30.5 on the Noometry Index.
Which is cheaper, GPT-4 Turbo or MiniMax-M2.5?
MiniMax-M2.5 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; GPT-4 Turbo lists at $10 and $30.
Is GPT-4 Turbo or MiniMax-M2.5 better for coding?
MiniMax-M2.5 scores higher on coding benchmarks: 48.1 versus 33.8 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M2.5 does, with 205K tokens against 128K.
How many benchmarks do GPT-4 Turbo and MiniMax-M2.5 share?
18 benchmarks have published results for both models. GPT-4 Turbo has 36 scored results on Noometry and MiniMax-M2.5 has 33.