Qwen3 32B: clarified the 32,768-token native context and 131,072-token YaRN-extended limit

The OnAirToday AI Progress Tracker recorded a verified change to Alibaba's Qwen3 32B (version Qwen3-32B): Clarified the 32,768-token native context and 131,072-token YaRN-extended limit.

Alibaba's Qwen3-32B is an open-weight, 32.8-billion-parameter text model with switchable thinking and non-thinking modes, multilingual support, and tool-calling capabilities. It remains a distinct release within the Qwen3 family, which also includes newer variants such as Qwen3-235B-A22B-Instruct-2507.

What changed in this review:

• Vendor thinking-mode results include GPQA Diamond at 68.4%, AIME 2024 at 81.4%, and AIME 2025 at 72.9%.
• API pricing depends on the hosting provider; no single model-wide input or output rate applies.
• Newer Qwen3-235B-A22B-Instruct-2507 is available; this profile remains specific to Qwen3-32B.
• Flaws and limitations updated
• Risks and threats updated

Teams building on Qwen3 32B 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/qwen3-32b.

References:

• https://huggingface.co/Qwen/Qwen3-32B
• https://qwenlm.github.io/blog/qwen3/
• https://aider.chat/docs/leaderboards/#polyglot-leaderboard
• https://huggingface.co/Qwen/Qwen3-235B-A22B-Instruct-2507