OpenAI launched Codex agent for software dev —Built on codex-1, an o3 variant fine-tuned for software engineering
—Runs in isolated clouds to write features, fix bugs, answer code questions, and run tests
—Available to ChatGPT Pro, Enterprise, Team users
LLMS
-
OpenAI Launches Codex Agent for Software Development Tasks
By
–
-
LLMs as Causal Agents: Decision-Making and Goal Generation
By
–
LLMs have causal power; they can be used to make decisions and control systems in the real world. LLMs can emulate agents, they can follow goals and generate goals to follow. LLMs can create a chain of thought reflect on it. LLMs generate and use abstractions. Perception is not
-
First Look at Video Overviews by NotebookLM
By
–
BREAKING 🚨: First look into Video Overviews, which are generated by the same model that will power NotebookLM soon.
— 🚨 AI News | TestingCatalog (@testingcatalog) 18 mai 2025
These are 4 "Sparks", 1-3 minute videos in different styles generated from various sources. pic.twitter.com/OgO1hDoP9mBREAKING : First look into Overviews, which are generated by the same model that will power NotebookLM soon. These are 4 "Sparks", 1-3 minute videos in different styles generated from various sources.
-
Zero-to-LLM Engineer Bundle: Complete Learning Path
By
–
Since many of you asked, here it is… our "Zero-to-LLM Engineer" bundle! We ( @towards_AI ) recently released our three core offers for any builders out there, together taking you from "zero" (literally, no coding knowledge or anything required) up to an advanced LLM developer
-
Neural Networks Outperform N-gram Models in Mixture Architecture
By
–
their neural networks worked better than any other methods and the bigger ones are better although at this scale, mixing with ngrams still helps a lot. they use a "mixture of models" – similar to today's MoEs, but the experts are different ngram models, plus one neural network
-
Model Training Scale: From 800K to 25 Trillion Tokens
By
–
their model is trained on 800K words, which is around 2.5M tokens. remember today's models are up to ~25T tokens (10^7 more, or 10 million million times) (they trained another model on 15M words)
-
Compact Language Model Architecture: 12M Parameters vs LLAMA 4
By
–
they use word embeddings + single-layer MLPs vocab size is 18K, context window is 6 words, hidden dimension is 60, word embedding have 100 dimensions their model has approx |V|(nm + h) = 17,964 × (6 × 100 + 60) = 12 million parameters about 200K times smaller than LLAMA 4…
-
Codex Model Limitations in Code Generation and Cursor Integration
By
–
My first experiences with Codex are that it makes the same mistakes other models do when one-shotting something in Cursor. It also failed to respond to feedback and hallucinated a bunch of stuff. Ended up fixing PR in Cursor. OpenAI also re-implemented a bunch of Github stuff
-
Negative Understanding of LLMs: A Gary Marcus Critique
By
–
to garymarcus harder than the man himself, you had to acquire not zero but negative understanding of LLMs; respect
-
Smart Terminal Assistant: Natural Language to Cross-Platform Commands
By
–
Smart Terminal Assistant A natural language interface that converts English to terminal commands across operating systems. Built with LangGraph's multi-agent system using A2A and MCP protocols for cross-platform execution. Check out the implementation