Model comparison
Kimi K2.6 vs MiniMax-M3
Kimi K2.6 is the stronger model overall, scoring 47.7 to 43.8 on the Noometry Index. MiniMax-M3 costs 3.3× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Last verified . 36 shared benchmarks.
MiniMaxSummary
- They share 36 benchmarks with published results for both. Kimi K2.6 scores higher in 7 categories and MiniMax-M3 in 3 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Kimi K2.6 leads 57.0 to 40.0.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 96.1% for Kimi K2.6 and 71.1% for MiniMax-M3.
- MiniMax-M3 is cheaper at $0.30 / $1.20 per million input/output tokens, against $0.95 / $4 for Kimi K2.6.
- MiniMax-M3 accepts more context: 1M tokens versus 262K.
Side by side
| Kimi K2.6 | MiniMax-M3 | |
|---|---|---|
| Provider | Moonshot AI | MiniMax |
| Noometry Index | 47.7 | 43.8 |
| Released | 2026-04-20 | 2026-06-01 |
| Weights | Open | Open |
| Context window | 262K | 1M |
| Max output | 262K | 512K |
| Input $ / M tokens | $0.95 | $0.30 |
| Output $ / M tokens | $4 | $1.20 |
| Results tracked | 51 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Kimi K2.6 leads
Kimi K2.6: 50.7 (#43), MiniMax-M3: 41.8 (#118)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| LMArena WebDev | 1509 | 1482 |
| SciCode | 53.5% | 47.1% |
| LMArena Coding | 1488 | 1469 |
| ALE-Bench | 1,093 | 640.02 |
| SWE-bench Verified | 76.7% | — |
| FrontierCode | — | 14.7% |
| WeirdML | 55.9% | — |
Agentic & Tool Use Too close to call
Kimi K2.6: 21.9 (#137), MiniMax-M3: 22.6 (#130)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| OSWorld 2.0 | 4.6% | 4.6% |
| GBAEval | 0.9% | 0.9% |
| Vending-Bench 2 | 6,205 | 2,158 |
| APEX-Agents | — | 37.7% |
| ExploitBench | 18.4% | — |
| GDP.pdf | 12% | — |
Reasoning Kimi K2.6 leads
Kimi K2.6: 40.5 (#55), MiniMax-M3: 30.1 (#87)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| NYT Connections (extended) | 87.2% | 65.1% |
| CritPt | 8% | 3.7% |
| Chess Puzzles | 26% | 14% |
| LMArena Hard Prompts | 1470 | 1447 |
| Mystery Game Puzzles | 18% | 8% |
| DTBench | 90.9% | 78.9% |
| LMCA | 37.3% | 33.7% |
| Epoch Capabilities Index | 151.05 | 146.95 |
| SimpleBench | — | 45.8% |
| EBR-Bench | 2.4% | — |
| Surface Evolver Bench | — | 55% |
| ForecastBench | — | 61.4 |
Math Kimi K2.6 leads
Kimi K2.6: 57.0 (#41), MiniMax-M3: 40.0 (#95)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 96.1% | 71.1% |
| ProofBench | 16% | 18% |
| LMArena Math | 1475 | 1429 |
| FrontierMath (Tiers 1-3) | 57.2% | — |
| FrontierMath Tier 4 | 25.6% | — |
| MathArena Final-Answer Competitions | 72.9% | — |
| FrontierMath (Feb 2025 set) | 39% | — |
| FrontierMath Tier 4 (v1) | 14.6% | — |
Knowledge MiniMax-M3 leads
Kimi K2.6: 54.0 (#54), MiniMax-M3: 58.4 (#35)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| GPQA Diamond | 90.8% | 90.9% |
| LMArena Expert | 1491 | 1461 |
| SimpleQA Verified | 34.9% | — |
| Vectara Hallucination Rate | 10.8% | — |
Multimodal MiniMax-M3 leads
Kimi K2.6: 31.6 (#103), MiniMax-M3: 40.2 (#51)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| LMArena Vision | 1283 | 1253 |
| LMArena Document | 1451 | 1435 |
| Blueprint-Bench 2 | 3.9% | — |
| Furniture Assembly | 21.7% | — |
Multilingual Kimi K2.6 leads
Kimi K2.6: 54.9 (#37), MiniMax-M3: 53.0 (#75)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| LMArena Non-English | 1446 | 1420 |
| LMArena Chinese | 1521 | 1463 |
| LMArena French | 1471 | 1447 |
| LMArena German | 1450 | 1426 |
| LMArena Japanese | 1443 | 1381 |
| LMArena Korean | 1427 | 1372 |
| LMArena Russian | 1446 | 1428 |
| LMArena Spanish | 1464 | 1432 |
Instruction Following Too close to call
Kimi K2.6: 76.3 (#43), MiniMax-M3: 75.5 (#62)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| LMArena Instruction Following | 1451 | 1433 |
Long Context Too close to call
Kimi K2.6: 44.9 (#52), MiniMax-M3: 44.2 (#72)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| LMArena Longer Query | 1468 | 1445 |
Writing & Preference Kimi K2.6 leads
Kimi K2.6: 68.5 (#26), MiniMax-M3: 62.1 (#83)
| Benchmark | Kimi K2.6 | MiniMax-M3 |
|---|---|---|
| LMArena Text | 1455 | 1433 |
| LMArena Creative Writing | 1434 | 1404 |
| EQ-Bench 4 | 1202 | 1150 |
| LMArena Multi-Turn | 1453 | 1442 |
| EQ-Bench Creative Writing | 1725 | — |
Frequently asked questions
Is Kimi K2.6 better than MiniMax-M3?
Kimi K2.6 is the stronger model overall, scoring 47.7 to 43.8 on the Noometry Index. MiniMax-M3 costs 3.3× less per token, which makes it the better buy when Kimi K2.6's lead doesn't matter for your workload.
Which is cheaper, Kimi K2.6 or MiniMax-M3?
MiniMax-M3 is cheaper. It lists at $0.30 per million input tokens and $1.20 per million output tokens; Kimi K2.6 lists at $0.95 and $4.
Is Kimi K2.6 or MiniMax-M3 better for coding?
Kimi K2.6 scores higher on coding benchmarks: 50.7 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 K2.6 and MiniMax-M3 share?
36 benchmarks have published results for both models. Kimi K2.6 has 51 scored results on Noometry and MiniMax-M3 has 41.