Model comparison
Kimi K2.6 vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 47.7 on the Noometry Index. Kimi K2.6 costs 1.8× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Last verified . 35 shared benchmarks.
Alibaba (Qwen)Summary
- They share 35 benchmarks with published results for both. Kimi K2.6 scores higher in 1 category and Qwen3.8 Max in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Qwen3.8 Max leads 45.4 to 21.9.
- The biggest single-benchmark swing is ProofBench: 16% for Kimi K2.6 and 58% for Qwen3.8 Max.
- Kimi K2.6 is cheaper at $0.95 / $4 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
- Qwen3.8 Max accepts more context: 1M tokens versus 262K.
- Kimi K2.6 has downloadable open weights; the other is API-only.
Side by side
| Kimi K2.6 | Qwen3.8 Max | |
|---|---|---|
| Provider | Moonshot AI | Alibaba (Qwen) |
| Noometry Index | 47.7 | 56.8 |
| Released | 2026-04-20 | 2026-08-02 |
| Weights | Open | Proprietary |
| Context window | 262K | 1M |
| Max output | 262K | 131K |
| Input $ / M tokens | $0.95 | $2 |
| Output $ / M tokens | $4 | $6 |
| Results tracked | 51 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
Kimi K2.6: 50.7 (#43), Qwen3.8 Max: 53.5 (#29)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| LMArena WebDev | 1509 | 1674 |
| SciCode | 53.5% | 53.2% |
| LMArena Coding | 1488 | 1502 |
| SWE-bench Verified | 76.7% | — |
| DeepSWE | — | 57.5% |
| FrontierSWE | — | 17.8% |
| WeirdML | 55.9% | — |
| ALE-Bench | 1,093 | — |
Agentic & Tool Use Qwen3.8 Max leads
Kimi K2.6: 21.9 (#137), Qwen3.8 Max: 45.4 (#14)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| GDP.pdf | 12% | 23.2% |
| APEX-Agents | — | 63.3% |
| OSWorld 2.0 | 4.6% | — |
| τ²-bench Banking | — | 55.1% |
| ExploitBench | 18.4% | — |
| GBAEval | 0.9% | — |
| Vending-Bench 2 | 6,205 | — |
Reasoning Qwen3.8 Max leads
Kimi K2.6: 40.5 (#55), Qwen3.8 Max: 54.4 (#26)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| NYT Connections (extended) | 87.2% | 88.3% |
| CritPt | 8% | 20% |
| Chess Puzzles | 26% | 40% |
| LMArena Hard Prompts | 1470 | 1496 |
| Mystery Game Puzzles | 18% | 38% |
| DTBench | 90.9% | 92% |
| LMCA | 37.3% | 46.2% |
| Epoch Capabilities Index | 151.05 | 156.41 |
| EBR-Bench | 2.4% | — |
Math Qwen3.8 Max leads
Kimi K2.6: 57.0 (#41), Qwen3.8 Max: 73.2 (#20)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| FrontierMath (Tiers 1-3) | 57.2% | 74.7% |
| FrontierMath Tier 4 | 25.6% | 46.3% |
| OTIS Mock AIME 2024-2025 | 96.1% | 100% |
| ProofBench | 16% | 58% |
| LMArena Math | 1475 | 1499 |
| MathArena Final-Answer Competitions | 72.9% | — |
| FrontierMath (Feb 2025 set) | 39% | — |
| FrontierMath Tier 4 (v1) | 14.6% | — |
Knowledge Qwen3.8 Max leads
Kimi K2.6: 54.0 (#54), Qwen3.8 Max: 61.7 (#27)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 90.8% | 92.7% |
| SimpleQA Verified | 34.9% | 47.3% |
| LMArena Expert | 1491 | 1507 |
| Vectara Hallucination Rate | 10.8% | — |
Multimodal Qwen3.8 Max leads
Kimi K2.6: 31.6 (#103), Qwen3.8 Max: 37.2 (#75)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | 1283 | 1314 |
| Furniture Assembly | 21.7% | 20% |
| Blueprint-Bench 2 | 3.9% | — |
| LMArena Document | 1451 | — |
Multilingual Qwen3.8 Max leads
Kimi K2.6: 54.9 (#37), Qwen3.8 Max: 56.7 (#18)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1446 | 1472 |
| LMArena Chinese | 1521 | 1538 |
| LMArena French | 1471 | 1503 |
| LMArena German | 1450 | 1483 |
| LMArena Japanese | 1443 | 1467 |
| LMArena Korean | 1427 | 1461 |
| LMArena Russian | 1446 | 1481 |
| LMArena Spanish | 1464 | 1492 |
Instruction Following Qwen3.8 Max leads
Kimi K2.6: 76.3 (#43), Qwen3.8 Max: 77.6 (#17)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1451 | 1479 |
Long Context Too close to call
Kimi K2.6: 44.9 (#52), Qwen3.8 Max: 45.6 (#31)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1468 | 1489 |
Writing & Preference Kimi K2.6 leads
Kimi K2.6: 68.5 (#26), Qwen3.8 Max: 67.1 (#30)
| Benchmark | Kimi K2.6 | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1455 | 1483 |
| LMArena Creative Writing | 1434 | 1479 |
| LMArena Multi-Turn | 1453 | 1489 |
| EQ-Bench Creative Writing | 1725 | — |
| EQ-Bench 4 | 1202 | — |
Frequently asked questions
Is Kimi K2.6 better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 47.7 on the Noometry Index. Kimi K2.6 costs 1.8× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Which is cheaper, Kimi K2.6 or Qwen3.8 Max?
Kimi K2.6 is cheaper. It lists at $0.95 per million input tokens and $4 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is Kimi K2.6 or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 50.7 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 Max does, with 1M tokens against 262K.
How many benchmarks do Kimi K2.6 and Qwen3.8 Max share?
35 benchmarks have published results for both models. Kimi K2.6 has 51 scored results on Noometry and Qwen3.8 Max has 39.