Qwen3 235B A22B Thinking: benchmark labels explicitly identify AIME 2025 and LiveCodeBench v6
OnAirToday’s tracker verifies the benchmark names against Alibaba documentation
Benchmark labels explicitly identify AIME 2025 and LiveCodeBench v6 — verified by the OnAirToday AI Progress Tracker against Alibaba's published documentation.
OnAirToday
Newsroom
The OnAirToday AI Progress Tracker recorded a verified change to Alibaba's Qwen3 235B A22B Thinking (version Qwen3-235B-A22B-Thinking-2507): Benchmark labels explicitly identify AIME 2025 and LiveCodeBench v6.
Qwen3-235B-A22B-Thinking-2507 is Alibaba's text-only, thinking-focused mixture-of-experts model, with 235 billion total parameters and 22 billion activated per token. Its published evaluations emphasize mathematical reasoning, scientific knowledge and coding, with a native 262,144-token context window and no switchable non-thinking mode.
What changed in this review:
- Capability score 64 → 73
- Flaws and limitations updated
Teams building on Qwen3 235B A22B Thinking should confirm limits, pricing and benchmark coverage against the vendor's own documentation before relying on these figures.
The full profile — capabilities, benchmark citations, known flaws and risk guidance — is on the tracker at /ai-models/qwen3-235b-a22b-thinking.
References:
- https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507
- https://huggingface.co/Qwen/Qwen3-235B-A22B
Generated insight by OnAirToday AI
Explicit benchmark versions make evaluation claims easier to trace and compare. They also reduce ambiguity about which AIME and LiveCodeBench editions were used.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.