Skip to content
Alibaba (Qwen)

Qwen 3.6 Max Preview

Frontier

Alibaba (Qwen)Released on 2026-04-20

Alibaba's most powerful Qwen model to date, released April 20, 2026. Tops multiple coding and agent benchmarks including SWE-bench Pro, Terminal-Bench 2.0, SkillsBench, QwenClawBench, QwenWebBench and SciCode. Hosted proprietary model, preview only.

87
Overall Score

Voice of the community

sample 45

Alibaba released Qwen 3.6-Max-Preview, topping SWE-bench Pro and multiple coding/agent benchmarks.

Decrypt2026-04-20

Core Specs

1000K
Context Window
0K
Max Output
ReasoningOpen Sourcetextimage

Pros & Cons

Sentiment70% +25% ·5% −

Pros

  • +#1 on SWE-bench Pro (tops all Claude/GPT/DeepSeek)
  • +Leader on multiple agent benchmarks
  • +1M-token context window
  • +Free trial access during preview

Cons

  • Preview — API pricing not set
  • Proprietary (not open-source)
  • Limited independent benchmark verification
  • Most benchmark #1 claims are from Alibaba

Pricing

Input (per 1M tokens)$0.00
Output (per 1M tokens)$0.00
Free trial available
Updated on 2026-04-21

Get Started

1Visit the provider's website
2Create an account
3Start using the model

Benchmarks

sweBenchPro#1 (top coding leaderboard)%
terminalBench2#1%
skillsBench#1%
qwenClawBench#1%
qwenWebBench#1%
sciCode#1%
noteRanked #1 on multiple coding and agent benchmarks per Alibaba announcement.%

Reliability

Incidents (30d)0
View Status Page →