QwQ-32B

Training compute

3.5×10²⁴ FLOP

Parameters

32.5B

Published

Mar 6, 2025

QwQ-32B is an AI model developed by Alibaba (China), first published in March 2025. It works in the language domain, on tasks such as language modeling/generation, question answering, quantitative reasoning and code generation.

Training it took an estimated 3.5×10²⁴ FLOP of compute. The model has 32,500,000,000 parameters.

Access: Open weights (unrestricted). Its weights are openly available. It is built on top of Qwen2.5-Coder (32B). Epoch AI rates the confidence of this record as speculative.

Full record

Organization: Alibaba
Country of organization: China
Domain: Language
Task: Language modeling/generation, Question answering, Quantitative reasoning, Code generation
Training compute: 3.5×10²⁴ FLOP
Parameters: 32,500,000,000
Numerical format: BF16
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Base model: Qwen2.5-Coder (32B)
Epoch confidence: Speculative

Benchmark results

01	LiveBench	Global average	72
02	Aider Polyglot	Percent correct	20.9