DeepCode: Open Agentic Coding Li et al.: https://
arxiv.org/abs/2512.07921 #ArtificialIntelligence #DeepLearning #MachineLearning
OPEN SOURCE
-
DeepCode: Open Agentic Coding System by Li et al.
By
–
-
Macaron achieves 1T parameter RL training at 10% cost
By
–
Every AI lab is pissing away millions on compute. Macaron just ran RL on a 1 trillion parameter model at 10% cost.
Then open sourced the whole damn thing. Already integrated into NVIDIA Megatron and ByteDance Seed Verl while most companies are still writing Medium posts about -
2.6K Parameter RL System Matches 62B Specialized Model
By
–
Oh My! I just built a new RL-native system with 2.6k parameters that matches a specialized 62B model from a 2024 paper, specifically on GSM8k. That's over 20,000,000x smaller than the baseline — but the approach is so different the traditional comparison doesn't make sense? 😉
-
AI Perfectly Captures Human Emotion in Swaps
By
–
Cast Swap: any face. Original emotion stays locked in. Scene Swap: different character, different world. Performance untouched. Every micro expression preserved. Every blink. Every breath. AI finally learned to capture human emotion instead of faking it.
-
First LLM Contact from Space Using Open Source Gemma Models
By
–
First LLM contact from space using our highly efficient open source Gemma models! Huge congrats to @PhilipJohnston and the @starcloud_ team!
-
MCP Joins Linux Foundation’s Agentic AI Initiative
By
–
MCP is now a part of the Agentic AI Foundation, a directed fund under the Linux Foundation.
— Anthropic (@AnthropicAI) 11 décembre 2025
Co-creator David Soria Parra talks about how a protocol sketched in a London conference room became the open standard for connecting AI to the world—and what comes next for it. pic.twitter.com/nIYPc15b4YMCP is now a part of the Agentic AI Foundation, a directed fund under the Linux Foundation. Co-creator David Soria Parra talks about how a protocol sketched in a London conference room became the open standard for connecting AI to the world—and what comes next for it.
-
GPT-5.2 Rollout and SOTA Performance
By
–
BREAKING : GPT-5.2 Instant, Thinking, and Pro are rolling out on ChatGPT. Free and Go users will receive it a bit later. Also, GPT-5.2 Pro (High) is SOTA for ARC-AGI-2, scoring 54.2% for $15.72/task! Big at coding and math
-
Building Autonomous Agents with aisuite Open Source Package
By
–
Sharing a fun recipe for building a highly autonomous, moderately capable, and very UNreliable agent using the open source aisuite package that Rohit Prasad and I have been working on. With a few lines of code, you can give a frontier LLM a tool (like disk access or web search),
-
GPT-5.2 Codename ‘Robin’ Confirmed
By
–
BREAKING : "Robin" is confirmed to be a codename for GPT-5.2 with a recent PR to Codex. Soon https://
x.com/synthwavedd/st
/synthwavedd/status/1999151342766391331
… -
Google Tests New Models: Fiercefalcon vs Ghostfalcon
By
–
Google is testing 2 new models on LM Arena "fiercefalcon" and "ghostfalcon". Potentially, these are either Gemini 3 Pro GA or Gemini 3 Flash. Who will have the best model by the end of 2025? https://
x.com/AiBattle_/stat
/AiBattle_/status/1998925023785128394
…