Alibaba25 Jul 2025

Qwen3-235B-A22B-Thinking-2507

Benchmark results

Benchmark	Score	Source check
Humanity's Last ExamNo tools · text-only subset	18.2%	Corroborated The same number is printed in 1 other lab document: 1
GPQA Diamond	81.1%	Corroborated The same number is printed in 2 other lab documents: 1 2
AIME 2025No tools	92.3%	Corroborated The same number is printed in 2 other lab documents: 1 2
MMLU-Proup to 32,768 output tokens	84.4%	Checked A person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
HMMT February 2025No tools · up to 81,920 output tokens	83.9%	Corroborated The same number is printed in 1 other lab document: 1
LiveCodeBenchLiveCodeBench v6, problems from Feb to May 2025	74.1%	Corroborated The same number is printed in 1 other lab document: 1
τ²-benchTelecom · up to 32,768 output tokens	45.6%	Checked A person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
τ²-benchRetail · up to 32,768 output tokens	71.9%	Checked A person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
τ²-benchAirline · up to 32,768 output tokens	58.0%	Checked A person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.

Sources

Reported by Alibaba