Telegraphing our @OpenAI model roadmap: • GPT-4.5 is next, our final non-chain-of-thought model
• GPT-5 will merge our GPT family and the o-series (including o3), enabling systems that can use all our tools and reasoning
LLMS
-
OpenAI Reveals GPT-4.5 and GPT-5 Model Roadmap Plans
By
–
-
Attention in Transformers: Concepts and Code PyTorch
By
–
New short course: Attention in Transformers: Concepts and Code in PyTorch.
— Andrew Ng (@AndrewYNg) 12 février 2025
Last week we released a course on how LLM transformers work. This week, go deeper and learn about the technical ideas behind the attention mechanism, and see how to code it in PyTorch. This course is… pic.twitter.com/nIeCFNz6PyNew short course: Attention in Transformers: Concepts and Code in PyTorch. Last week we released a course on how LLM transformers work. This week, go deeper and learn about the technical ideas behind the attention mechanism, and see how to code it in PyTorch. This course is
-
Sam Altman Says AI Will Soon Auto-Prompt Itself
By
–
Comme je l'avais annoncé : cela ne sert à rien de se former à l'intelligence artificielle puisque l'IA va bientôt s'auto-prompter. C'est ce que Sam Altman vient d'annoncer. Interagir avec l'IA sera aussi simple qu'une conversation et nous donnera les réponses d'un prompt avancé
-
The Prompt Is Dead: A Told-You-So Moment
By
–
Voilà : le prompt est mort ! (I don't want to say I told you so, but I told you so!)
-
Two educators create attention mechanisms course collaboration
By
–
Whoa! It's great to see two of my favorite educators team up to create a course on attention mechanisms! (PS: If you found Ch 3 of my Build a Large Language Model From Scratch book a bit dry, this looks like a great supplementary resource!)
-
OpenAI Roadmap Update: GPT-4.5 and GPT-5 Product Simplification
By
–
OPENAI ROADMAP UPDATE FOR GPT-4.5 and GPT-5: We want to do a better job of sharing our intended roadmap, and a much better job simplifying our product offerings. We want AI to “just work” for you; we realize how complicated our model and product offerings have gotten. We hate
-
LangChain Launch: New AI Integration Available Today
By
–
Excited to see this launch! Try it out in LangChain today
-
Stanford Releases Sub-$50 AI Reasoning Model, Commoditizing Advanced Models
By
–
Stanford scholars including @drfeifei
, @percyliang
, and @tatsu_hashimoto released a paper last week on an AI “reasoning” model trained for under $50, raising real questions about the commoditization of AI models. Read via @TechCrunch
: -
Multi-Agent System Architectures: Supervisor Patterns Released
By
–
There are many ways to build multi-agent systems – we just released a prebuilt package for one of the more common ones (supervisor/hierarchical) What common patterns/architectures should we do next?
-
Prompt Injection Attempts with Invisible Bytes in AI Models
By
–
I'm able to do basic prompt injections with the invisible bytes but I can't get it to work without explicit decoding hints. https://
chatgpt.com/share/67acd3ba
-d234-8007-ad44-ba9d4dfc2920
… The thinking models actually feel a bit more susceptible because they love puzzles and they notice the added bytes and get very