Model comparison
Qwen3.6 Max Preview vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 51.5 on the Noometry Index.
Last verified . 24 shared benchmarks.
Alibaba (Qwen)Summary
- They share 24 benchmarks with published results for both. Qwen3.6 Max Preview scores higher in 0 categories and Qwen3.8 Max in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where Qwen3.8 Max leads 73.2 to 54.1.
- The biggest single-benchmark swing is Chess Puzzles: 20% for Qwen3.6 Max Preview and 40% for Qwen3.8 Max.
- Both cost about the same: $1.30 input and $7.80 output per million tokens.
- Qwen3.8 Max accepts more context: 1M tokens versus 262K.
Side by side
| Qwen3.6 Max Preview | Qwen3.8 Max | |
|---|---|---|
| Provider | Alibaba (Qwen) | Alibaba (Qwen) |
| Noometry Index | 51.5 | 56.8 |
| Released | 2026-04-20 | 2026-08-02 |
| Weights | Proprietary | Proprietary |
| Context window | 262K | 1M |
| Max output | 66K | 131K |
| Input $ / M tokens | $1.30 | $2 |
| Output $ / M tokens | $7.80 | $6 |
| Results tracked | 29 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
Qwen3.6 Max Preview: 48.7 (#54), Qwen3.8 Max: 53.5 (#29)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| LMArena WebDev | 1482 | 1674 |
| LMArena Coding | 1471 | 1502 |
| SWE-bench Verified | 76.7% | — |
| DeepSWE | — | 57.5% |
| FrontierSWE | — | 17.8% |
| SciCode | — | 53.2% |
Agentic & Tool Use Not comparable
Qwen3.6 Max Preview: —, Qwen3.8 Max: 45.4 (#14)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| APEX-Agents | — | 63.3% |
| τ²-bench Banking | — | 55.1% |
| GDP.pdf | — | 23.2% |
| Vending-Bench 2 | 4,254 | — |
Reasoning Qwen3.8 Max leads
Qwen3.6 Max Preview: 41.7 (#53), Qwen3.8 Max: 54.4 (#26)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| NYT Connections (extended) | 74.1% | 88.3% |
| Chess Puzzles | 20% | 40% |
| LMArena Hard Prompts | 1457 | 1496 |
| Mystery Game Puzzles | 19% | 38% |
| DTBench | 87.2% | 92% |
| LMCA | 42.5% | 46.2% |
| Epoch Capabilities Index | 149.24 | 156.41 |
| SimpleBench | 63% | — |
| CritPt | — | 20% |
Math Qwen3.8 Max leads
Qwen3.6 Max Preview: 54.1 (#46), Qwen3.8 Max: 73.2 (#20)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 91.1% | 100% |
| LMArena Math | 1465 | 1499 |
| FrontierMath (Tiers 1-3) | — | 74.7% |
| FrontierMath Tier 4 | — | 46.3% |
| ProofBench | — | 58% |
| FrontierMath (Feb 2025 set) | 23.1% | — |
| FrontierMath Tier 4 (v1) | 4.2% | — |
Knowledge Qwen3.8 Max leads
Qwen3.6 Max Preview: 57.6 (#39), Qwen3.8 Max: 61.7 (#27)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 87.4% | 92.7% |
| SimpleQA Verified | 52% | 47.3% |
| LMArena Expert | 1478 | 1507 |
Multimodal Not comparable
Qwen3.6 Max Preview: —, Qwen3.8 Max: 37.2 (#75)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | — | 1314 |
| Furniture Assembly | — | 20% |
Multilingual Qwen3.8 Max leads
Qwen3.6 Max Preview: 54.2 (#48), Qwen3.8 Max: 56.7 (#18)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1437 | 1472 |
| LMArena Chinese | 1487 | 1538 |
| LMArena French | 1449 | 1503 |
| LMArena Russian | 1445 | 1481 |
| LMArena Spanish | 1454 | 1492 |
| LMArena German | — | 1483 |
| LMArena Japanese | — | 1467 |
| LMArena Korean | — | 1461 |
Instruction Following Qwen3.8 Max leads
Qwen3.6 Max Preview: 75.7 (#55), Qwen3.8 Max: 77.6 (#17)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1438 | 1479 |
Long Context Qwen3.8 Max leads
Qwen3.6 Max Preview: 44.6 (#61), Qwen3.8 Max: 45.6 (#31)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1457 | 1489 |
Writing & Preference Qwen3.8 Max leads
Qwen3.6 Max Preview: 63.8 (#60), Qwen3.8 Max: 67.1 (#30)
| Benchmark | Qwen3.6 Max Preview | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1447 | 1483 |
| LMArena Creative Writing | 1435 | 1479 |
| LMArena Multi-Turn | 1456 | 1489 |
Frequently asked questions
Is Qwen3.6 Max Preview better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 51.5 on the Noometry Index.
Which is cheaper, Qwen3.6 Max Preview or Qwen3.8 Max?
Qwen3.6 Max Preview is cheaper. It lists at $1.30 per million input tokens and $7.80 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is Qwen3.6 Max Preview or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 48.7 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 Max does, with 1M tokens against 262K.
How many benchmarks do Qwen3.6 Max Preview and Qwen3.8 Max share?
24 benchmarks have published results for both models. Qwen3.6 Max Preview has 29 scored results on Noometry and Qwen3.8 Max has 39.