My Favorites This Week
OpenAI o3-mini
Jan 31, 2025
OpenAI unveiled two new powerful and fast reasoning models: o3-mini and o3-mini-high. These models excel in math, coding, and science.
DeepSeek Janus-Pro-7B
Jan 27, 2025
DeepSeek unveiled a new open-source AI model named Janus-Pro-7B. This multimodal LLM possesses the ability to comprehend visual information and generate images.
huggingface deepseek-ai/Janus-Pro-7B · Hugging Face
Alibaba Qwen2.5-Max
Jan 28, 2025
Qwen2.5-Max is an AI model that can perform various tasks, including working with code, writing text, searching the web, generating images, and even creating videos.
Qwen Qwen Team Qwen2.5-Max: Exploring the Intelligence of Large-scale MoE Model
Kimi Reasoning Model
Jan 26, 2025
Kimi unveiled its Kimi k1.5 multimodal AI model, which is comparable to OpenAI's o1 model, having a context of 128k and the ability to process both text and visual inputs.
Other Releases
Google AI Weather Forecast
Jan 31, 2025
Google unveiled its most cutting-edge AI models for weather forecasting, called WeatherNext. These models significantly enhanced disaster response capabilities and allowed the seamless integration of real-time and historical data, accelerating research efforts.
Krea AI Chat
Jan 30, 2025
This new tool integrates every Krea feature into a chat interface using DeepSeek.
Mistral Small 3
Jan 30, 2025
Mistral Small 3 is a pre-trained and instruction-following model designed to handle tasks which demand language comprehension and adherence to instructions while maintaining low latency. It is comparable to Llama 3.3 70B and Qwen 32B, and it has the potential to replace GPT4o-mini.
Pika 2.1 Video Generation Model
Jan 28, 2025
Pika released the newest model called Pika 2.1. It comes with 1080p resolution and more details, seamless motion, and lifelike human characters.
DeepSeek R1 in Perplexity
Jan 27, 2025
Perplexity now enables deep web research using the powerful DeepSeek reasoning model.
👏 Thanks for reading!