Check out the NinjaChat AI platform over here : https://www.ninjachat.ai/
USE COUPON CODE "KING25" for 25% OFF on ALL MEMBERSHIPS ON ninjachat.ai
In this video, I'll be telling you about Qwen QwQ-32B Preview that claims to beat OpenAI's O1 & Sonnet Model. This is a new model and can do reasoning and is fully opensource. Today, I'll be testing it and check if it can really beat Sonnet and O1 or not.
----
Resources:
QwQ 32B : https://ollama.com/library/qwq
----
Key Takeaways:
🎉 Qwen's New QwQ Model is Here! - Qwen has introduced the powerful QwQ-32B reasoning model, an open-source AI that rivals top models like O1 and DeepSeek.
🚀 Run Locally with Ease! - This 32b AI model can run on GPUs like the NVIDIA 4090, bringing cutting-edge AI capabilities to your personal setup.
📊 Performance Insights! - QwQ-32B shines in benchmarks like AIME and Math 500, though it still trails in GPQA and LiveCodeBench.
🔀 Mind the Limitations. - While it’s promising, issues like language mixing, code-switching, and circular reasoning show there’s room for growth.
💡 Perfect for Developers! - Open weights, a demo, and compatibility with Ollama make it a fantastic choice for testing and integrating into projects.
🖥️ Hands-On Testing in Action! - We tested QwQ on 13 creative questions, showcasing its abilities in coding, reasoning, and more – with a 70% accuracy compared to O1.
🌟 Open Source is the Future! - With models like QwQ, AI technology is becoming more accessible and exciting. Imagine what’s next as open-source AI evolves!
----
Timestamps:
00:00 - Introduction
00:51 - NinjaChat (Sponsor)
01:59 - About Qwen QwQ 32B Model
03:40 - Testing
09:14 - Final Results, Charts & Opinions
09:52 - Ending
