Tech News
GPT-6 Sol and Luna
OpenAI releases GPT-6 Sol and Luna, with Luna priced at half the cost of GPT-5.6 Luna. Early users highlight model-specific behaviors, including pelican generation comparisons and agentic engineering instincts.
Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war
Yesterday was Grok 4.7 ( pelicans ) and MiMo v2.6 Flash/Pro ( more pelicans ). Today Anthropic released Claude Opus 5.5, and around an hour later OpenAI released GPT-6 Sol and GPT-6 Luna.
vllm-project/vllm v0.30.0
v0.30.0 Highlights This release features 762 commits from 315 contributors (104 new)!
How Trane gets building insights 60x faster with Amazon Bedrock AgentCore
In about four weeks, Trane Technologies built an AI-powered agentic solution on Amazon Bedrock AgentCore that reduced a 20-minute, multi-screen building diagnostic workflow to a 20-second natural language...
Parallel cut research time and cost in half with GPT‑6 Astra
GPT‑6 Astra allowed Parallel’s agents to research and synthesize labor-market data in half the time and at half the cost.
GitHub Repos
HKUDS/nanobot
Ultra-lightweight, open-source, self-hosted personal AI agent framework in Python with WebUI, tools, memory, MCP, multi-agent workflows, automation, and chat apps
# Python# agent framework# ai agentComposioHQ/awesome-claude-skills
A curated list of awesome Claude Skills, resources, and tools for customizing Claude AI workflows
# Python# agent skills# ai agentsgeneralaction/emdash
Emdash is the Open-Source Agentic Development Environment (🧡 YC W26).
# TypeScript# agenticdevelopment# agenticdevelopmentenvironmentResearch Papers
RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
This paper proposes RRSI, a method that regularizes the automated evolution of LLM agent harnesses to reduce overfitting to training tasks. It constrains how edits are proposed and selected, leading to better out-of-distribution performance and fewer tokens used.
When Quantization Preserves Accuracy but Not Evidence: Explanation-Aware Post-Training Quantization for Medical LLMs
This paper shows that quantizing medical LLMs can keep multiple-choice accuracy high while making the model's explanations less faithful. It proposes a quantization method that tries to preserve the evidence supporting answers, not just the final answer.
Toward a foundation model for forest point clouds
These findings identify the practical regime in which pretrained representations are most valuable and suggest that instance discrimination, rather than forest semantics, is the main remaining obstacle to a general-purpose 3D forest foundation model.