LLM post-training
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.
Nous Releases Arete-27B to Alter Qwen’s Political Framing
Nous releases Arete-27B, a Qwen3.8-27B fine-tune targeting Chinese-state framing. Its bias and capability results come from Nous’s own tests.
Latest
Together AI Discloses Prompt Leakage After a Dragos Study
Together AI discloses prompt leakage of stored samples after a public credential exposure. OpenRouter found no breach of itself; Novita cites a routing issue.
Apple Releases FastFrames Open Video-Generation Models
Apple has released FastFrames, research video models that the paper says reach interactive latency, under licences that are not unrestricted open source.
Microsoft Unveils MAI Code 2, Cyber 2 Flash and Perception
Microsoft has launched MAI Code 2, Code 2 Flash and Cyber 2 Flash at Ignite, and said its Perception security system is now generally available.
OpenAI Launches GPT-6.2 Astra and Sol, Now on Bedrock
OpenAI has launched GPT-6.2 Astra and GPT-6.2 Sol in ChatGPT and the API, with same-day general availability on Amazon Bedrock.
Replit Launches Missions, Agent Swarms With Spending Caps
Replit has launched Missions, which it says runs agent swarms toward a goal within a budget, with spending caps and approve-before-publishing controls.
Cognition Launches Devin Bots to Lead Teams of Agents
Cognition has launched Devin Bots, persistent agents it says can lead teams of Devins and bring engineers only the decisions that need their judgment.
Mistral Launches Vibe Desktop for macOS and Windows
Mistral has launched Vibe Desktop for macOS and Windows, combining Work and coding in one app. Linux is coming soon; no separate CLI install is required.
AMD Completes Its All-Stock Acquisition of World Labs
AMD has completed its acquisition of World Labs, with Fei-Fei Li joining as chief scientist. The all-stock deal was valued at about $8.2bn in September.
Tencent Adds Hy4 Flash to TokenHub After the Hy4 Launch
Tencent has added Hy4 flash to TokenHub. Specs, prices and test results were reported by TechNode, not on Tencent’s public docs.
Alibaba Releases Qwen4-Max With a One-Million-Token Window
Alibaba has released Qwen4-Max, a 2.4-trillion-parameter flagship model with a one-million-token context window, on QwenCloud and its API.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.
