GPT-56T 861 —
MUSE-SPK 837 +0.2%
GPT-56SC 790 -4.6%
GLM-5 781 -0.4%
CL-OP55X 780 -5.1%
GROK-46H 780 -5.1%
QWEN-38X 748 -9.2%
GPT-6A 743 -9.4%
KIMI-K3X 742 —
CL-FAB5H 698 -6.1%
CL-OP5H 675 -6.2%
GEM-38FH 672 -0.7%
CL-OP5X 670 -5.5%
CL-OP55H 668 —
CL-OP46H 657 -5.9%
CL-OP47H 648 -6.1%
GPT-56S 618 -0.6%
GEM-37FH 610 -7.2%
GEM-36FH 593 —
CL-OP48H 588 —
CL-OP47 581 -0.2%
GEM-35FH 580 —
GPT-55H 541 -7%
INKL 531 —
GEM-31P 512 -0.2%
CL-OP46 498 +0.4%
GEM-3P 498 -0.2%
CL-OP48 492 +0.4%
GPT-52 464 —
GPT-55 423 —
GPT-56T 861 —
MUSE-SPK 837 +0.2%
GPT-56SC 790 -4.6%
GLM-5 781 -0.4%
CL-OP55X 780 -5.1%
GROK-46H 780 -5.1%
QWEN-38X 748 -9.2%
GPT-6A 743 -9.4%
KIMI-K3X 742 —
CL-FAB5H 698 -6.1%
CL-OP5H 675 -6.2%
GEM-38FH 672 -0.7%
CL-OP5X 670 -5.5%
CL-OP55H 668 —
CL-OP46H 657 -5.9%
CL-OP47H 648 -6.1%
GPT-56S 618 -0.6%
GEM-37FH 610 -7.2%
GEM-36FH 593 —
CL-OP48H 588 —
CL-OP47 581 -0.2%
GEM-35FH 580 —
GPT-55H 541 -7%
INKL 531 —
GEM-31P 512 -0.2%
CL-OP46 498 +0.4%
GEM-3P 498 -0.2%
CL-OP48 492 +0.4%
GPT-52 464 —
GPT-55 423 —
← Back to feed

Alibaba Releases Qwen3.6 Max Preview Free — AA Score 52 Places It Atop Every Chinese Model

Alibaba released Qwen3.6 Max Preview on April 20, 2026, adding the top tier of its flagship model family. Third-party evaluator Artificial Analysis scored it 52 on its Intelligence Index, placing it #2 overall out of 201 tracked models and above every other Chinese model currently on the leaderboard — including GLM-5.1 and MiniMax M2.7.

The release completes the Qwen3.6 series. Alibaba now ships four distinct tiers: the new Max Preview (closed, free), Plus (closed, paid at 2 yuan/$0.28 per million input tokens), Flash (closed, fast tier), and the open-source 35B-A3B under Apache 2.0.

What Max Improves Over Plus

Alibaba published head-to-head results showing Max Preview leads Plus across six benchmarks: SWE-Bench Pro, Terminal-Bench 2.0, SkillsBench, QwenClawBench, QwenWebBench, and SciCode. Qwen3.6-Plus was already competitive — 61.6% on Terminal-Bench 2.0 (ahead of the then-current Claude Opus 4.5), 78.8% on SWE-Bench Verified, and 56.6% on SWE-Bench Pro. Max Preview improves on all six benchmarks without publishing exact scores at launch.

Beyond coding, Max Preview reports higher scores on world knowledge benchmarks (SuperGPQA, QwenChineseBench) and instruction following (ToolcallFormatIFBench). Alibaba describes the gains as “more reliable for complex, long-horizon agent tasks.”

Access and Pricing

During the preview period, Qwen3.6 Max Preview is available for free via Qwen Studio. API access is coming via Alibaba Cloud Bailian under the model string qwen3.6-max-preview. Pricing has not been announced.

The model carries a 256K-token context window. No multimodal input is listed at this time — the specification covers text-to-text.

Leaderboard Position

Artificial Analysis ranks Qwen3.6 Max Preview at #2 globally in its Intelligence Index version 4.0, which incorporates ten evaluations: GDPval-AA, τ²-Bench Telecom, Terminal-Bench Hard, SciCode, AA-LCR, AA-Omniscience, IFBench, Humanity’s Last Exam, GPQA Diamond, and CritPt. Only Claude Opus 4.7 (max) and Gemini 3.1 Pro Preview score higher on the AA leaderboard.

For context, Qwen3.6-Plus is currently at $0.276 per million input tokens on OpenRouter, ranks first by usage on OpenRouter’s daily, weekly, and trending charts, and Qwen3.6-35B-A3B tops the HuggingFace open-source leaderboard. The addition of a free flagship preview reinforces Alibaba’s pattern of flooding the market across all price tiers simultaneously.

Key Numbers

  • AA Intelligence Index: 52 (#2 / 201 models)
  • Context window: 256K tokens
  • Benchmarks improved over Plus: SWE-Bench Pro, Terminal-Bench 2.0, SkillsBench, QwenClawBench, QwenWebBench, SciCode
  • Price (preview): free via Qwen Studio
  • API model string: qwen3.6-max-preview (pricing TBD)