Superseded lineage (verified 2026-09-30)
QwQ-32B was Alibaba’s 2024-25 reasoning model. Reasoning moved into the main line with Qwen3 (April 2025, “thinking mode”). Per the Qwen Wikipedia page and the Qwen Hugging Face org: Qwen3.5 (Feb 2026), Qwen3.6 (Apr 2026, Apache), Qwen3.7-Max/Plus (proprietary), Qwen3.8 (Aug 2026):
Qwen3.8-2.4T-A95B(open weights 2026-08-12, commercial-restricted licence),Qwen3.8-27B(Apache 2.0, 2026-08-14),Qwen3.8-Flash-Next. Qwen 4 was announced at Apsara on 2026-09-22 but is unreleased (no date/specs). Provider overview: ai-model-providers.
QwQ is the reasoning model of the Qwen series
2025-03: 32B model on ollama (original note). Wikipedia dates the first QwQ release to November 2024 (preview); the March 2025 date here was not re-verified.
Sources
- https://en.wikipedia.org/wiki/Qwen (2026-09-30)
- https://huggingface.co/Qwen and https://huggingface.co/Qwen/Qwen3.8-2.4T-A95B (2026-09-30)
- https://www.cnbc.com/2026/08/17/alibaba-meta-qwen-open-weight-ai-laptop-models.html (search snippet)
Related
- Qwen family (current lineup and 2025 generations)