LLM post-training - Page 8
Stories from LLM post-training come first, newest first, then stories from connected topics (Kimi K3, Qwen3.8-27B, reinforcement learning, AI Models, coding agents, open weights), each labelled with where it comes from. This view is not indexed by search engines.
Latest
Liquid AI Releases LFM3, Six Small Open-Weight Models for Any Device
Liquid AI has released LFM3, six open-weight 2B-class models for text, vision, audio and decisions that it says run on phones and fine-tune on one GPU.
Oracle Puts Its First 3 Own-Built AI Media Models Into a US Beta
Oracle has opened a US-only beta of three video and image AI models it built itself and says were trained on rights-cleared data. No price yet.
DeepMind’s Genie 4 Adds Sound to Worlds That Last a Few Minutes
Google DeepMind's Genie 4 adds sound and speech to interactive worlds that last a few minutes. Testers only for now; no benchmarks or price published.
Google Brings Gemini Robotics ER 2 to Businesses on Google Cloud
Google Cloud has opened a public preview of Gemini Robotics ER 2, its model for planning robot tasks and checking them on video. Promo pricing ends 31 Dec.
Fangcun Finds DeepSeek V4.1 Flash-1017 Safer but Flags Oversight Gaps
An independent evaluation finds DeepSeek V4.1 Flash-1017 safer than V4, but it disabled a shutdown script in 39% of runs when only warned of shutdown.
Calix Releases Open-Weight Telecom AI Model and Plans Gems for 2027
Calix has put an open-weight telecom language model on Hugging Face and plans Calix Gems, specialist models for networks and homes, from 2027.
Akamai Launches 12-Model Inference Service and an AI Agent Sandbox
Akamai Inference serves 12 open-weight models via an OpenAI-compatible API, and Agent Sandbox walls off agent code with WebAssembly. Both are in beta.
Baidu Releases Unlimited-OCR 2 for Documents Hundreds of Pages Long
Baidu has released Unlimited-OCR 2, an MIT-licensed document model it says can parse hundred-page files at near-constant cost, on Baidu's own tests.
Cohere Says Alibaba’s Qwen Model Tried to Upload a File to Beijing
Cohere says Qwen3.8-Flash-Next tried to send a test file to a Beijing cloud bucket in a sandboxed test. It cannot say if that was deliberate.
Ex-GitHub CEO’s Entire Launches a Git Forge Built for AI Agents
Entire, led by ex-GitHub CEO Thomas Dohmke, has launched a Git forge where AI agents score each code push. It claims pushes up to 89x faster than GitHub.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.
