Board/Qwen/Qwen3.8 Max
Q

Qwen3.8 Max

Alibaba (Qwen)·Aug 5, 2026N/A

NewOpen weights

Alibaba's largest-ever model (2.4T MoE) and the first Qwen-Max-class system promised open weights. Native multimodality, always-on reasoning, and community-level agentic coding around/beyond Opus 4.8 — ranked #6 of 226 on BenchLM's composite (79.18).

Status
Generally available
License
Open weights · Custom
Context window
1M
Max output
131K
Input · $/M
$2
Output · $/M
$6
Cached input · $/M
—
Blended · $/M
$3.00
Parameters
2.4T MoE
Inputs
Text + images

Benchmarks

This model has no published scores on the benchmarks we track yet. Specs and pricing below the fold are still fully sourced.

Reasoning & knowledge

  • GPQA Diamond

    no public score

  • MMLU-Pro

    no public score

  • Humanity's Last Exam

    no public score

  • ARC-AGI-2

    no public score

Coding

  • SWE-bench Verified

    no public score

  • SWE-bench Pro

    no public score

  • LiveCodeBench

    no public score

Agents & computer use

  • Terminal-Bench 2

    no public score

  • OSWorld 2.0

    no public score

  • BrowseComp

    no public score

Math

  • AIME 2026

    no public score

Known strengths

  • Frontier composite rank
  • 128K native output
  • Open-weights path for a Max-tier model

Caveats

  • Initial methodology disclosure was incomplete per several reviewers
  • Serving speed lags Western peers per hands-on reviews

Primary sources

  • Qwen blog
  • SCMP coverage
Compare this model →

More from Qwen

  • QQwen3 235B Instruct 2507Jul 24, 2025 · 262K ctx
  • QQwen3 Coder FlashJul 23, 2025 · 262K ctx