Alibaba’s Qwen team has introduced QwQ-32B, a groundbreaking AI model with 32 billion parameters that demonstrates performance comparable to the larger DeepSeek-R1 model. This breakthrough showcases the potential of scaling Reinforcement Learning (RL) on robust foundation models, highlighting the effectiveness of this approach in enhancing model performance.

The Qwen team has successfully integrated agent capabilities into the reasoning model, enabling it to think critically, utilize tools, and adapt its reasoning based on environmental feedback. This integration is a significant step forward in developing more advanced AI models.

According to the team, “Scaling RL has the potential to enhance model performance beyond conventional pretraining and post-training methods. Recent studies have demonstrated that RL can significantly improve the reasoning capabilities of models.” This is evident in QwQ-32B’s performance, which rivals that of DeepSeek-R1, despite having significantly fewer parameters.

QwQ-32B has achieved impressive results in various benchmarks, including AIME24, LiveCodeBench, LiveBench, IFEval, and BFCL, designed to assess its mathematical reasoning, coding proficiency, and general problem-solving capabilities. Its performance is comparable to that of leading models, including DeepSeek-R1-Distilled-Qwen-32B, DeepSeek-R1-Distilled-Llama-70B, o1-mini, and the original DeepSeek-R1.

The benchmark results are as follows:

• AIME24: QwQ-32B achieved a score of 79.5, slightly behind DeepSeek-R1-6718’s 79.8, but significantly ahead of OpenAl-o1-mini’s 63.6 and the distilled models.
• LiveCodeBench: QwQ-32B scored 63.4, closely matched by DeepSeek-R1-6718’s 65.9, and surpassing the distilled models and OpenAl-o1-mini’s 53.8.
• LiveBench: QwQ-32B achieved a score of 73.1, with DeepSeek-R1-6718 scoring 71.6, and outperforming the distilled models and OpenAl-o1-mini’s 57.5.
• IFEval: QwQ-32B scored 83.9, very close to DeepSeek-R1-6718’s 83.3, and leading the distilled models and OpenAl-o1-mini’s 59.1.
• BFCL: QwQ-32B achieved a score of 66.4, with DeepSeek-R1-6718 scoring 62.8, demonstrating a lead over the distilled models and OpenAl-o1-mini’s 49.3.

The Qwen team’s approach involved a cold-start checkpoint and a multi-stage RL process driven by outcome-based rewards. The initial stage focused on scaling RL for math and coding tasks, utilizing accuracy verifiers and code execution servers. The second stage expanded to general capabilities, incorporating rewards from general reward models and rule-based verifiers.

According to the team, “We find that this stage of RL training with a small amount of steps can increase the performance of other general capabilities, such as instruction following, alignment with human preference, and agent performance, without significant performance drop in math and coding.” This approach has significant implications for the development of more advanced AI models.

QwQ-32B is open-weight and available on Hugging Face and ModelScope under the Apache 2.0 license, and is also accessible via Qwen Chat. The Qwen team views this as an initial step in scaling RL to enhance reasoning capabilities and aims to further explore the integration of agents with RL for long-horizon reasoning.

As the team stated, “As we work towards developing the next generation of Qwen, we are confident that combining stronger foundation models with RL powered by scaled computational resources will propel us closer to achieving Artificial General Intelligence (AGI).” This is an exciting development in the field of AI, with significant potential for future advancements.

See also: Deepgram Nova-3 Medical: AI speech model cuts healthcare transcription errors

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.