Salesforce has introduced Koa, an enterprise-focused language model created by post-training the open-weight Nemotron-3-Super-120B foundation model. The training process uses reinforcement learning with Group Relative Policy Optimization (GRPO), drawing on public data and other sources. The work targets agentic tool use in enterprise settings, according to the arXiv preprint.
A preprint on arXiv examines how segmenting context affects the performance of open-weight small language models used for capture-the-flag cybersecurity exercises. The authors frame the work around the risk that locally hosted models can sidestep the guardrails enforced by proprietary APIs. The paper is a revised cross-listing in the cs.AI category.
OpenAI has published the largest version of GPT-2, containing 1.5 billion parameters, as the last step of its staged release plan. The model weights and code are being made available to support research into detecting text generated by GPT-2 systems. The company notes that larger language models have appeared since its earlier staged releases, but it chose to complete the original release schedule.
WHY IT MATTERS ↘Completing the staged release effectively concedes that withholding weights offers little protection once comparable models are widely available, shifting the burden onto detection and misuse mitigation rather than access control. It also sets a precedent that release schedules, once announced, tend to be honored even as the frontier moves past them.