Model comparison
Grok 4.20 Multi-Agent vs MiniMax-M2.5
Grok 4.20 Multi-Agent is the stronger model overall, scoring 45.5 to 38.1 on the Noometry Index. MiniMax-M2.5 costs 3.0× less per token, which makes it the better buy when Grok 4.20 Multi-Agent's lead doesn't matter for your workload.
Last verified . 18 shared benchmarks.
MiniMaxSummary
- They share 18 benchmarks with published results for both. Grok 4.20 Multi-Agent scores higher in 7 categories and MiniMax-M2.5 in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Grok 4.20 Multi-Agent leads 42.4 to 17.6.
- The biggest single-benchmark swing is NYT Connections (extended): 89.6% for Grok 4.20 Multi-Agent and 16.8% for MiniMax-M2.5.
- MiniMax-M2.5 is cheaper at $0.30 / $1.20 per million input/output tokens, against $1.25 / $2.50 for Grok 4.20 Multi-Agent.
- Grok 4.20 Multi-Agent accepts more context: 1M tokens versus 205K.
- MiniMax-M2.5 has downloadable open weights; the other is API-only.
Side by side
| Grok 4.20 Multi-Agent | MiniMax-M2.5 | |
|---|---|---|
| Provider | xAI | MiniMax |
| Noometry Index | 45.5 | 38.1 |
| Released | 2026-03-09 | 2026-02-12 |
| Weights | Proprietary | Open |
| Context window | 1M | 205K |
| Max output | 30K | 131K |
| Input $ / M tokens | $1.25 | $0.30 |
| Output $ / M tokens | $2.50 | $1.20 |
| Results tracked | 20 | 33 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding MiniMax-M2.5 leads
Grok 4.20 Multi-Agent: 42.4 (#91), MiniMax-M2.5: 47.9 (#53)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Coding | 1458 | 1380 |
| SWE-bench Verified (bash only) | — | 75.8% |
| LMArena WebDev | — | 1386 |
| SWE-bench Multilingual | — | 68.3% |
| ALE-Bench | — | 618.17 |
Agentic & Tool Use Not comparable
Grok 4.20 Multi-Agent: —, MiniMax-M2.5: 30.2 (#70)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| Terminal-Bench | — | 42.7% |
| LMArena Search | 1204 | — |
| Vending-Bench 2 | — | -23.16 |
Reasoning Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 42.4 (#44), MiniMax-M2.5: 17.6 (#281)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| NYT Connections (extended) | 89.6% | 16.8% |
| LMArena Hard Prompts | 1448 | 1372 |
| ARC-AGI-2 | — | 4.9% |
| Kagi LLM Benchmark | — | 55.2% |
| ARC-AGI-1 | — | 63.7% |
| Epoch Capabilities Index | — | 146.68 |
Math Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 36.6 (#106), MiniMax-M2.5: 25.8 (#242)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Math | 1441 | 1378 |
| ProofBench | — | 4% |
Knowledge Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 39.7 (#116), MiniMax-M2.5: 38.6 (#132)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Expert | 1445 | 1378 |
| Vectara Hallucination Rate | — | 9.1% |
Multimodal Not comparable
Grok 4.20 Multi-Agent: 39.8 (#44), MiniMax-M2.5: —
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Vision | 1259 | — |
Multilingual Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 54.6 (#40), MiniMax-M2.5: 47.3 (#147)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Non-English | 1440 | 1338 |
| LMArena Chinese | 1475 | 1392 |
| LMArena French | 1467 | 1363 |
| LMArena German | 1456 | 1361 |
| LMArena Japanese | 1405 | 1169 |
| LMArena Korean | 1416 | 1232 |
| LMArena Russian | 1457 | 1359 |
| LMArena Spanish | 1447 | 1354 |
Instruction Following Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 75.0 (#80), MiniMax-M2.5: 71.7 (#143)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Instruction Following | 1420 | 1353 |
Long Context Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 43.7 (#84), MiniMax-M2.5: 37.4 (#211)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Longer Query | 1430 | 1367 |
| CL-bench | — | 11.4% |
| CL-bench Life | — | 6.3% |
Writing & Preference Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 64.1 (#57), MiniMax-M2.5: 54.1 (#149)
| Benchmark | Grok 4.20 Multi-Agent | MiniMax-M2.5 |
|---|---|---|
| LMArena Text | 1450 | 1359 |
| LMArena Creative Writing | 1435 | 1331 |
| LMArena Multi-Turn | 1451 | 1364 |
| EQ-Bench Creative Writing | — | 1361 |
Frequently asked questions
Is Grok 4.20 Multi-Agent better than MiniMax-M2.5?
Grok 4.20 Multi-Agent is the stronger model overall, scoring 45.5 to 38.1 on the Noometry Index. MiniMax-M2.5 costs 3.0× less per token, which makes it the better buy when Grok 4.20 Multi-Agent's lead doesn't matter for your workload.
Which is cheaper, Grok 4.20 Multi-Agent 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; Grok 4.20 Multi-Agent lists at $1.25 and $2.50.
Is Grok 4.20 Multi-Agent or MiniMax-M2.5 better for coding?
MiniMax-M2.5 scores higher on coding benchmarks: 47.9 versus 42.4 in the Noometry coding category.
Which has the bigger context window?
Grok 4.20 Multi-Agent does, with 1M tokens against 205K.
How many benchmarks do Grok 4.20 Multi-Agent and MiniMax-M2.5 share?
18 benchmarks have published results for both models. Grok 4.20 Multi-Agent has 20 scored results on Noometry and MiniMax-M2.5 has 33.