state of open-source AI in 2025:
– almost all new open American models are finetuned Chinese base models
– we don’t know the base models’ training data
– we have no idea how to audit or “decompile” base models who knows what could be hidden in the weights of DeepSeek
LLMS
-
Open-source AI transparency crisis: Chinese base models audit challenges
By
–
-
Groq’s Crazy October: Partnerships, Price Drops, and Major Updates
By
–
october was craaaazy:
– Groq IBM partnership
– McLaren keeps winning with Groq on the halo
– gpt-oss-safeguard day 0 on GroqCloud
– gpt-oss 120/20B major price drop + prompt caching
– Huge openbench v0.5.0 release with 350+ new evals
– @JonathanRoss321 speaks at @FIIKSA – Groq -
Attending EMNLP2025 in Suzhou presenting multimodal LLM grounding work
By
–
I will be heading to Suzhou soon attending #EMNLP2025 next week, presenting my work on grounding multimodal LLMs with world/cultural knowledge, https://
arxiv.org/abs/2508.07414. Email/DM open if you want to chat about the paper and other topics more broadly(multimodal LLMs x languages x -
Langrepl CLI: Terminal Tool for LLM Agents with Visual Debugging
By
–
Langrepl CLI (Made by the LangChain Community) A sophisticated terminal tool for building LLM agents with deep agents, featuring persistent conversations and LangGraph Studio integration for visual debugging. Key Features:
– Deep Agent Architecture
– Visual debugging -
AI Education: Using GPT5 as Your Personal Learning Tutor
By
–
Après avoir lu « Ne faites plus d’études » nos lecteurs demandent à @OlivierBabeau et moi Comment me former à l’IA ? La réponse est simple : demandez à GPT5 un plan de formation adapté à vos connaissances et à votre structure cognitive C’est le meilleur professeur au monde
-
Efficiency and Intelligence in AI Agent Development
By
–
By the end of testing, it was beating every open-source agent in the field and holding its own against proprietary systems built with 10x the resources. Turns out intelligence isn’t about size. It’s about how well you manage what you already know.
-
AI models demonstrating reflection and iterative problem-solving
By
–
It even knows when it’s wasting time. After 11 failed attempts in one example, it folded the entire dead end into one sentence, learned the lesson, and moved on. That’s not just optimization that’s reflection.
-
Breakthrough in Scalable Reasoning Across Model Depth
By
–
Here’s where it gets wild other models plateau after 64 steps. This one keeps getting better all the way up to 256 turns. It learns across depth, not just width the first real sign of scalable reasoning.
-
Optimizing AI Agent Context Management and Token Efficiency
By
–
Most agents collapse when their context fills up.
This one doesn’t. Even after 100 turns, its entire memory grows from ~3.5K to just ~7K tokens. That’s like running a 500-page research project and still remembering only what matters. And instead of stacking logs endlessly like -
Autonomous AI Agent Logic for Note Curation and Information Compression
By
–
Instead of drowning in its own notes, the agent actively curates them. Every turn it asks: what’s worth keeping, and what can I compress? It runs a loop that looks almost human think → fold → explain → act and builds its own hierarchy of ideas.