Reflecting back, these were the biggest technical lessons for me in AI in the past five years: 2020: you can cast any language task as sequence prediction and learn it via pretrain + finetune 2021: scaling to GPT-3 size enables doing arbitrary tasks specified via instructions
LLMS
-
OpenAI Upcoming Features: AGI, Agents, and Advanced Capabilities
By
–
common themes: AGI
agents
much better 4o upgrade
much better memory
longer context
“grown up mode”
deep research feature
better sora
more personalization (interestingly, many great updates we have coming were mentioned not at all or very little!) -
Meta Research on Generative Retrieval for Recommendation Systems
By
–
Newly published research for generative retrieval for recommendations from teams at Meta. – Preference Discerning with LLM-Enhanced Generative Retrieval https://
go.fb.me/evvcu8
– Unifying Generative and Dense Retrieval for Sequential Recommendation https://
go.fb.me/i7l955 -
Best ChatGPT Prompt Infographic of 2024
By
–
Best CHATGPT Prompt Infographic of 2024 Source: https://
superhuman.ai -
New Models Coming Soon and Rate Limits Improvements Expected
By
–
And yes you can expect new models and continuous work on improving rate limits (thank you for bearing with us rn ) so let's get those out of the way to start
-
Anthropic’s 2025 Plans: What Would You Like to See?
By
–
Very excited for what's already in store for next year but there's always more we can do What would you like to see Anthropic build/fix in 2025?
-
Reasoning Models Transform AI Prompting in 2025
By
–
Prompting models in 2025 is going to be totally different. Reasoning models are going to completely change the way we use AI. Instead of having to craft the best prompt, we are starting to see the emergence of AI "self-prompting" – especially with reasoning models like o1. Why
-
DeepSeek Breaks Into Top 10 With Minimal Post-Training Cost
By
–
DeepSeek COOKED! It is the only open-weight (commercially permissive) model in the top 10. More so, it is a model that has been post-trained on ~5K H800 GPU hours (0.01M $) *only* – now imagine what a better post-trained model can do!
-
Qwen QwQ and 2.5 Coder Enter Top AI Models Ranking
By
–
QwQ + 2.5 Coder is there – The rest didn't make the hype list haha! – Lots of work to do in 2025 to get all Qwen models in top 25.