Qwen Unveils QwQ-32B: Efficient AI for Local Use; OpenAI Rolls Out ChatGPT 4.5
New AI Models
Release of QwQ-32B by Qwen:
Qwen has released QwQ-32B, a medium-sized reasoning AI model that can run on a single RTX 3090 GPU using 5-bit quantization. The model is noted for its efficiency and impressive benchmark results, making it accessible for local use.
- Better than Deepseek, New QwQ-32B, Thanx Qwen
- QwQ-32B released, equivalent or surpassing full Deepseek-R1!
- Qwen/QwQ-32B · Hugging Face
- Are we ready!
AMD Releases Instella 3B Model:
AMD has introduced the Instella 3B model, a fully open-source AI model designed to run on AMD hardware, supporting both NPU and GPU execution.
OpenAI Rolls Out ChatGPT 4.5:
OpenAI has started rolling out ChatGPT 4.5 to Plus users, with a rate limit of 50 messages per week. The rollout is expected to take 1-3 days, and users have reported mixed experiences with the new model.
- 4.5 Rolling out to Plus users
- GPT-4.5 is officially rolling out to Plus users!
- 4.5 Rollout to Plus Users Complete
AI Agents and Services
OpenAI Introduces New AI Agents:
OpenAI is planning to introduce three types of agents with pricing ranging from $2,000 to $20,000 per month. The most advanced agents, aimed at PhD-level tasks, will be priced at $20,000 per month. These agents are expected to offer significant productivity gains, potentially replacing multiple human roles and working 24/7 without breaks.
Manus.im AI Agent from China:
Manus.im, a new AI agent from China, has reportedly surpassed OpenAI's Deep Research model in the GAIA benchmarks, achieving state-of-the-art results. The agent is described as fully autonomous, capable of executing complex tasks and adaptive learning.
Experimental Models
Qwen Team Releases Qwq:
The Qwen Team has released Qwq, a 32B parameter experimental research model focused on advancing AI reasoning capabilities. The model is available on Ollama's platform, with users discussing its performance and comparing it to other models like llama.cpp.