Qwen3.6-Max-PreviewvsQwen3-Max-Thinking
Across 7 shared benchmarks, Qwen3.6-Max-Preview leads overall: Qwen3.6-Max-Preview wins 6, Qwen3-Max-Thinking wins 1, with 0 ties and an average score difference of +2.01.
Qwen3.6-Max-Preview
阿里巴巴 · 2026-04-18 · Chat model
Qwen3-Max-Thinking
阿里巴巴 · 2026-01-26 · Reasoning model
Qwen3.6-Max-Preview6 wins(86%)(14%)1 winQwen3-Max-Thinking
Benchmark scores
Grouped by capability, sorted by largest gap within each. 7 shared benchmarks.
Coding and Software Engineer
Qwen3.6-Max-Preview 2/2| Benchmark | Qwen3.6-Max-Preview | Qwen3-Max-Thinking | Diff |
|---|---|---|---|
| SWE-bench Verified | 78.8021 / 114Thinking (With Tools) | 75.3038 / 114 | +3.50 |
| LiveCodeBench | 87.1012 / 126最高（无工具） | 85.9016 / 126 | +1.20 |
General Knowledge
Qwen3.6-Max-Preview 2/2| Benchmark | Qwen3.6-Max-Preview | Qwen3-Max-Thinking | Diff |
|---|---|---|---|
| MMLU Pro | 88.505 / 133最高（无工具） | 85.7023 / 133 | +2.80 |
| HLE | 50.2029 / 181Thinking (With Tools) | 49.8032 / 181 | +0.40 |
General Evaluation
Qwen3.6-Max-Preview 1/1| Benchmark | Qwen3.6-Max-Preview | Qwen3-Max-Thinking | Diff |
|---|---|---|---|
| GPQA Diamond | 90.4036 / 226最高（无工具） | 87.4066 / 226 | +3 |
Instruction Following
Qwen3.6-Max-Preview 1/1| Benchmark | Qwen3.6-Max-Preview | Qwen3-Max-Thinking | Diff |
|---|---|---|---|
| IF Bench | 74.209 / 33最高（无工具） | 70.9014 / 33 | +3.30 |
Math and Reasoning
Qwen3-Max-Thinking 1/1| Benchmark | Qwen3.6-Max-Preview | Qwen3-Max-Thinking | Diff |
|---|---|---|---|
| IMO-AnswerBench | 83.8014 / 23最高（无工具） | 83.9013 / 23 | -0.10 |
Specs
| Field | Qwen3.6-Max-Preview | Qwen3-Max-Thinking |
|---|---|---|
| Publisher | 阿里巴巴 | 阿里巴巴 |
| Release date | 2026-04-18 | 2026-01-26 |
| Model type | Chat model | Reasoning model |
| Architecture | Dense | MoE |
| Parameters | Not available | 1T |
| Context length | 262K | 1000K |
| Max output | 64K | 32K |
API pricing
Prices use DataLearner records when available; missing fields are not inferred.
| Item | Qwen3.6-Max-Preview | Qwen3-Max-Thinking |
|---|---|---|
| Text input | $1.3 / 1M tokens | $1.2 / 1M tokens |
| Text output | $7.8 / 1M tokens | $6 / 1M tokens |
Summary
- Qwen3.6-Max-Previewleads in:Coding and Software Engineer (2/2), General Knowledge (2/2), General Evaluation (1/1), Instruction Following (1/1)
- Qwen3-Max-Thinkingleads in:Math and Reasoning (1/1)
On average across the 7 shared benchmarks, Qwen3.6-Max-Preview scores 2.01 higher.
Largest single-benchmark gap: SWE-bench Verified — Qwen3.6-Max-Preview 78.80 vs Qwen3-Max-Thinking 75.30 (+3.50).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.