Qwen3 32B: newer Qwen3-2507 family checkpoints shipped in July 2025; this profile remains specific to Qwen3-32B

The OnAirToday AI Progress Tracker recorded a verified change to Alibaba's Qwen3 32B (version Qwen3-32B): Newer Qwen3-2507 family checkpoints shipped in July 2025; this profile remains specific to Qwen3-32B.

Qwen3-32B is Alibaba's open-weight dense text model, combining switchable thinking and non-thinking modes with multilingual and tool-use capabilities. Newer Qwen3-2507 family checkpoints shipped in July 2025, but they are separate releases rather than updates to this specific checkpoint.

What changed in this review:

• Alibaba reports GPQA Diamond at 68.4% in thinking mode, rather than 65.7%.
• Published thinking-mode results include AIME 2024, AIME 2025 and LiveCodeBench v5.
• Hosted inference pricing is provider-specific; no provider-independent token prices are assigned.
• Flaws and limitations 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://arxiv.org/abs/2505.09388
• https://huggingface.co/Qwen/Qwen3-235B-A22B-Instruct-2507