Tech News
Deno Desktop
Deno Desktop allows building desktop apps with Deno, using a shared runtime to reduce binary sizes, and integrates with Tauri for cross-platform deployment.
The text in Claude Code’s “Extended Thinking” output
Claude Code's 'Extended Thinking' output is not the model's actual reasoning but a summary, raising concerns about transparency in AI models from major companies like Anthropic, OpenAI, and Google.
GLM 5.2 vs. Opus
A comparison of GLM 5.2 against Claude Opus highlights that while GLM 5.2 is a major step up from other non-frontier models, it still lags behind Opus. The discussion criticizes one-shot prompting benchmarks as unreal.
Show HN: Oak – Git alternative designed for agents
Oak is a new version control system designed for AI agents, offering virtual mounts to avoid full repo downloads and enable parallel tasking without conflicts. It's early-stage, bootstrapped on itself, but lacks Windo.
SpaceX sheds $400B in market value as debut rally hits reverse
SpaceX's market value has dropped by $400B from its peak, reversing its post-IPO rally. The article and comments question the sustainability of its valuation and the synergies between Musk's ventures.
GitHub Repos
open-multi-agent/open-multi-agent
TypeScript multi-agent orchestration framework.
# TypeScript# agent framework# agent orchestrationhiyouga/LlamaFactory
Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)
# Python# agent# aiUnicomAI/wanwu
China Unicom's Yuanjing Wanwu Agent Platform is an enterprise-grade, multi-tenant AI agent development platform.
# Go# agent# agentic aiResearch Papers
OpenCUA: Open Foundations for Computer-Use Agents
OpenCUA provides an open-source framework for building computer-use agents, including a large dataset (AgentNet) and models that achieve state-of-the-art performance among open-source models on OSWorld benchmarks.
Diversity-Aware Policy Optimization for Large Language Model Reasoning
This paper shows that promoting solution diversity during reinforcement learning training improves LLM reasoning performance, and proposes a token-level diversity objective that yields a 3.5% average improvement on math benchmarks.
AIR: Adaptive Interleaved Reasoning with Code in MLLMs
Following the paradigm shift initiated by OpenAI o3, interleaved reasoning with code to enhance multimodal large language models (MLLMs) has become a pivotal research frontier. The existing literature focuses primarily on tool-use within vision-perception tasks.