Qwen3 235B A22B: separate Qwen3-235B-A22B-Instruct-2507 and Thinking-2507 successors shipped in July 2025; this profile remains about the original release

The OnAirToday AI Progress Tracker recorded a verified change to Alibaba's Qwen3 235B A22B (version qwen3-235b-a22b): Separate Qwen3-235B-A22B-Instruct-2507 and Thinking-2507 successors shipped in July 2025; this profile remains about the original release.

Qwen3-235B-A22B is Alibaba's open-weight, text-based mixture-of-experts model, with 235 billion total parameters and 22 billion activated per token. This original release supports switchable thinking and non-thinking modes; separate Instruct-2507 and Thinking-2507 successors have since shipped.

What changed in this review:

• Context documentation distinguishes 32,768 native tokens from 131,072 tokens with YaRN.
• The original vendor evaluation reports GPQA Diamond at 71.1% and LiveCodeBench at 70.7%.
• Published AIME results are 85.7% for 2024 and 81.5% for 2025.
• Flaws and limitations updated
• Risks and threats updated

Teams building on Qwen3 235B A22B 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.

References:

• https://huggingface.co/Qwen/Qwen3-235B-A22B
• https://qwenlm.github.io/blog/qwen3/
• https://huggingface.co/Qwen/Qwen3-235B-A22B-Instruct-2507
• https://huggingface.co/Qwen/Qwen3-235B-A22B-Thinking-2507