Model comparison
Llama-3.3-70B-Instruct vs MiniMax-M2.1
MiniMax-M2.1 is the stronger model overall, scoring 38.9 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 3.4× 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.
Meta
MiniMaxSummary
- They share 18 benchmarks with published results for both. Llama-3.3-70B-Instruct scores higher in 0 categories and MiniMax-M2.1 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where MiniMax-M2.1 leads 38.3 to 15.3.
- The biggest single-benchmark swing is Vectara Hallucination Rate: 4.1% for Llama-3.3-70B-Instruct and 11.8% for MiniMax-M2.1.
- Llama-3.3-70B-Instruct is cheaper at $0.10 / $0.32 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
| Llama-3.3-70B-Instruct | MiniMax-M2.1 | |
|---|---|---|
| Provider | Meta | MiniMax |
| Noometry Index | 30.6 | 38.9 |
| Released | 2024-12-06 | 2025-12-23 |
| Weights | Open | Open |
| Context window | 128K | 205K |
| Max output | 4K | 131K |
| Input $ / M tokens | $0.10 | $0.30 |
| Output $ / M tokens | $0.32 | $1.20 |
| Results tracked | 43 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 31.0 (#290), MiniMax-M2.1: 40.4 (#143)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Coding | 1268 | 1421 |
| LMArena WebDev | — | 1384 |
| SciCode | 26% | — |
| WeirdML | 14.4% | — |
| BigCodeBench Instruct | 46.9% | — |
| LiveBench Coding | 36.6% | — |
| BigCodeBench Complete | 57.5% | — |
| ALE-Bench | — | 623.83 |
Agentic & Tool Use MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 25.8 (#105), MiniMax-M2.1: 27.9 (#98)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| Terminal-Bench | — | 36.6% |
| Berkeley Function Calling Leaderboard | 31.9% | — |
| BALROG | 23% | — |
Reasoning MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 14.1 (#327), MiniMax-M2.1: 16.6 (#302)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Hard Prompts | 1257 | 1411 |
| SimpleBench | 19.9% | — |
| NYT Connections (extended) | — | 11.2% |
| CritPt | 0% | — |
| LiveBench Reasoning | 50.8% | — |
| DTBench | 59.5% | — |
| LiveBench Data Analysis | 49.5% | — |
| LMCA | 17.5% | — |
| Epoch Capabilities Index | 127.33 | — |
| ForecastBench | 58.6 | — |
| LiveBench | 50.2% | — |
Math MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 15.3 (#298), MiniMax-M2.1: 38.3 (#138)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Math | 1267 | 1397 |
| OTIS Mock AIME 2024-2025 | 5.1% | — |
| LiveBench Math | 42.2% | — |
| MATH Level 5 | 41.6% | — |
Knowledge MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 30.6 (#226), MiniMax-M2.1: 38.3 (#147)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| Vectara Hallucination Rate | 4.1% | 11.8% |
| LMArena Expert | 1225 | 1431 |
| GPQA Diamond | 47.4% | — |
| Confabulations | 22.8% | — |
| MMLU | 86.3% | — |
Multilingual MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 39.9 (#220), MiniMax-M2.1: 50.0 (#128)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Non-English | 1236 | 1378 |
| LMArena Chinese | 1217 | 1430 |
| LMArena French | 1281 | 1404 |
| LMArena German | 1251 | 1381 |
| LMArena Japanese | 1150 | 1287 |
| LMArena Korean | 1143 | 1298 |
| LMArena Russian | 1252 | 1387 |
| LMArena Spanish | 1270 | 1397 |
Instruction Following MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 71.1 (#157), MiniMax-M2.1: 73.8 (#112)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Instruction Following | 1242 | 1400 |
| LiveBench Instruction Following | 82.7% | — |
Long Context MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 26.4 (#295), MiniMax-M2.1: 43.2 (#101)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Longer Query | 1256 | 1416 |
| Fiction.LiveBench | 33.3% | — |
Writing & Preference MiniMax-M2.1 leads
Llama-3.3-70B-Instruct: 47.6 (#207), MiniMax-M2.1: 58.3 (#120)
| Benchmark | Llama-3.3-70B-Instruct | MiniMax-M2.1 |
|---|---|---|
| LMArena Text | 1274 | 1392 |
| LMArena Creative Writing | 1250 | 1361 |
| LMArena Multi-Turn | 1280 | 1396 |
| LiveBench Language | 39.2% | — |
Frequently asked questions
Is Llama-3.3-70B-Instruct better than MiniMax-M2.1?
MiniMax-M2.1 is the stronger model overall, scoring 38.9 to 30.6 on the Noometry Index. Llama-3.3-70B-Instruct costs 3.4× less per token, which makes it the better buy when MiniMax-M2.1's lead doesn't matter for your workload.
Which is cheaper, Llama-3.3-70B-Instruct or MiniMax-M2.1?
Llama-3.3-70B-Instruct is cheaper. It lists at $0.10 per million input tokens and $0.32 per million output tokens; MiniMax-M2.1 lists at $0.30 and $1.20.
Is Llama-3.3-70B-Instruct or MiniMax-M2.1 better for coding?
MiniMax-M2.1 scores higher on coding benchmarks: 40.4 versus 31.0 in the Noometry coding category.
Which has the bigger context window?
MiniMax-M2.1 does, with 205K tokens against 128K.
How many benchmarks do Llama-3.3-70B-Instruct and MiniMax-M2.1 share?
18 benchmarks have published results for both models. Llama-3.3-70B-Instruct has 43 scored results on Noometry and MiniMax-M2.1 has 22.