# Kimi K2.5 vs MiniMax-M3

> Kimi K2.5 is the stronger model overall, scoring 48.1 to 43.8 on the Noometry Index. MiniMax-M3 costs 1.7× less per token, which makes it the better buy when Kimi K2.5's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/kimi-k2-5-vs-minimax-m3
- Last updated: 2026-10-10
- Shared benchmarks: 30

## Summary

- They share 30 benchmarks with published results for both. Kimi K2.5 scores higher in 8 categories and MiniMax-M3 in 2 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.5 leads 51.8 to 40.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 92.2% for Kimi K2.5 and 71.1% for MiniMax-M3.
- MiniMax-M3 is cheaper at $0.30 / $1.20 per million input/output tokens, against $0.45 / $2.25 for Kimi K2.5.
- MiniMax-M3 accepts more context: 1M tokens versus 262K.

## Snapshot

| | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| Provider | Moonshot AI | MiniMax |
| Noometry Index | 48.1 | 43.8 |
| Rank | 57 | 85 |
| Context | 262K | 1M |
| Input $/M | $0.45 | $0.30 |
| Output $/M | $2.25 | $1.20 |
| Weights | Open | Open |

## Coding

- Kimi K2.5: 48.8 (#53)
- MiniMax-M3: 41.8 (#118)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| LMArena WebDev | 1437 | 1482 |
| SciCode | 49% | 47.1% |
| LMArena Coding | 1474 | 1469 |
| ALE-Bench | 821.65 | 640.02 |
| SWE-bench Verified | 73.8% | — |
| FrontierCode | — | 14.7% |
| SWE-bench Verified (bash only) | 70.8% | — |
| SWE-bench Multilingual | 67.3% | — |
| WeirdML | 45.6% | — |

## Agentic & Tool Use

- Kimi K2.5: 34.2 (#48)
- MiniMax-M3: 22.6 (#130)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| Vending-Bench 2 | 1,198 | 2,158 |
| Terminal-Bench | 43.2% | — |
| APEX-Agents | — | 37.7% |
| OSWorld 2.0 | — | 4.6% |
| OSWorld | 63.3% | — |
| GBAEval | — | 0.9% |

## Reasoning

- Kimi K2.5: 31.2 (#80)
- MiniMax-M3: 30.1 (#87)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| SimpleBench | 46.8% | 45.8% |
| NYT Connections (extended) | 69.9% | 65.1% |
| CritPt | 3.1% | 3.7% |
| Chess Puzzles | 12% | 14% |
| LMArena Hard Prompts | 1453 | 1447 |
| Epoch Capabilities Index | 148.03 | 146.95 |
| ARC-AGI-2 | 11.8% | — |
| Kagi LLM Benchmark | 78.5% | — |
| ARC-AGI-1 | 65.3% | — |
| EnigmaEval | 3.4% | — |
| Thematic Generalization | 69.4% | — |
| Mystery Game Puzzles | — | 8% |
| DTBench | — | 78.9% |
| LMCA | — | 33.7% |
| Surface Evolver Bench | — | 55% |
| ForecastBench | — | 61.4 |

## Math

- Kimi K2.5: 51.8 (#53)
- MiniMax-M3: 40.0 (#95)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 92.2% | 71.1% |
| LMArena Math | 1470 | 1429 |
| MathArena Final-Answer Competitions | 62.3% | — |
| ProofBench | — | 18% |
| FrontierMath (Feb 2025 set) | 27.9% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |

## Knowledge

- Kimi K2.5: 53.6 (#56)
- MiniMax-M3: 58.4 (#35)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| GPQA Diamond | 87.6% | 90.9% |
| LMArena Expert | 1466 | 1461 |
| Humanity's Last Exam | 24.4% | — |
| SimpleQA Verified | 34.3% | — |
| Vectara Hallucination Rate | 14.2% | — |

## Multimodal

- Kimi K2.5: 41.1 (#39)
- MiniMax-M3: 40.2 (#51)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| LMArena Vision | 1269 | 1253 |
| LMArena Document | 1430 | 1435 |

## Multilingual

- Kimi K2.5: 53.9 (#53)
- MiniMax-M3: 53.0 (#75)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| LMArena Non-English | 1433 | 1420 |
| LMArena Chinese | 1495 | 1463 |
| LMArena French | 1454 | 1447 |
| LMArena German | 1441 | 1426 |
| LMArena Japanese | 1421 | 1381 |
| LMArena Korean | 1410 | 1372 |
| LMArena Russian | 1435 | 1428 |
| LMArena Spanish | 1450 | 1432 |

## Instruction Following

- Kimi K2.5: 75.3 (#64)
- MiniMax-M3: 75.5 (#62)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| LMArena Instruction Following | 1431 | 1433 |

## Long Context

- Kimi K2.5: 52.1 (#7)
- MiniMax-M3: 44.2 (#72)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| LMArena Longer Query | 1445 | 1445 |
| Fiction.LiveBench | 86.1% | — |
| CL-bench | 19.3% | — |
| CL-bench Life | 13.2% | — |

## Writing & Preference

- Kimi K2.5: 65.1 (#53)
- MiniMax-M3: 62.1 (#83)

| Benchmark | Kimi K2.5 | MiniMax-M3 |
|---|---|---|
| LMArena Text | 1445 | 1433 |
| LMArena Creative Writing | 1423 | 1404 |
| LMArena Multi-Turn | 1444 | 1442 |
| EQ-Bench Creative Writing | 1579 | — |
| EQ-Bench 4 | — | 1150 |

## FAQ

### Is Kimi K2.5 better than MiniMax-M3?

Kimi K2.5 is the stronger model overall, scoring 48.1 to 43.8 on the Noometry Index. MiniMax-M3 costs 1.7× less per token, which makes it the better buy when Kimi K2.5's lead doesn't matter for your workload.

### Which is cheaper, Kimi K2.5 or MiniMax-M3?

MiniMax-M3 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; Kimi K2.5 lists at $0.45 and $2.25.

### Is Kimi K2.5 or MiniMax-M3 better for coding?

Kimi K2.5 scores higher on coding benchmarks: 48.8 versus 41.8 in the Noometry coding category.

### Which has the bigger context window?

MiniMax-M3 does, with 1M tokens against 262K.

### How many benchmarks do Kimi K2.5 and MiniMax-M3 share?

30 benchmarks have published results for both models. Kimi K2.5 has 51 scored results on Noometry and MiniMax-M3 has 41.