Eon Systems Emulates Fruit Fly Brain as Karpathy Unveils Self-Improving Agentic Swarms

Neurotechnology & Biological AI

Eon Systems Achieves Full Neural Emulation of a Fruit Fly: Scientists have successfully created a neuron-for-neuron digital replica of a fruit fly's brain within a computer simulation. This digital entity demonstrated autonomous behaviors such as walking, grooming, and feeding, marking a major milestone in embodied brain emulation and biomechanical modeling.

• Scientists at Eon Systems just copied a fruit fly's brain into a computer. Neuron by neuron. It started walking, grooming, and feeding, doing what flies do all on its own

AI Agents & Optimization

Andrej Karpathy Launches Autonomously Improving Agentic Swarm: Andrej Karpathy has developed an operational agentic swarm designed to optimize AI models without human intervention. The system successfully discovered 20 architectural changes that improved validation loss, resulting in an 11% reduction in training time for a nanochat model.

• Andrej Karpathy's Newest Development - Autonomously Improving Agentic Swarm Is Now Operational

Large Language Models

Release of Uncensored Qwen3.5 Model Series: A new suite of Qwen3.5 models, ranging from 4B to 35B parameters, has been released in GGUF format with all refusals and safety filters removed. These "Aggressive" versions maintain the original model's multimodal capabilities while providing an unrestricted experience for local deployment.

• Qwen3.5-35B-A3B Uncensored (Aggressive) — GGUF Release
  • https://huggingface.co/HauhauCS/Qwen3.5-35B-A3B-Uncensored-HauhauCS-Aggressive
  • https://huggingface.co/HauhauCS/Qwen3.5-4B-Uncensored-HauhauCS-Aggressive
  • https://huggingface.co/HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive
  • https://huggingface.co/HauhauCS/Qwen3.5-27B-Uncensored-HauhauCS-Aggressive
  • http://huggingface.co/HauhauCS/models

Comparison Highlights Intelligence Gaps Between GPT-4.1 and GPT-5.3: New comparisons between GPT-4.1 and GPT-5.3 Chat suggest that the former may offer higher raw intelligence and speed for specific tasks. While GPT-5.3 is the version utilized within the ChatGPT interface, GPT-4.1 is noted for its distinct reasoning capabilities and processing efficiency.

• Is GPT-4.1 a smarter model than GPT-5.3 Chat?

AI Hardware & Benchmarking

Apple M5 Max AI Performance Benchmarks Revealed: Initial benchmarks for the newly arrived M5 Max laptop demonstrate its ability to run large-scale local models like Qwen3.5-122B and GPT-OSS-120B. The results provide detailed metrics on prompt processing and token generation speeds, showcasing the chip's high-end memory bandwidth and processing power.

• M5 Max just arrived - benchmarks incoming