News
AI News Brief — May 12
GitHub velocity is led by openai/codex; paper attention is clustering around LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling; social attention is tilting toward Ask GN: What do you think about services that involve human intervention in an environment wher…; biggest mover: affaan-m/everything-claude-code (+9). 10 repo signals, 10 paper picks, and 10 community items made today's cut.
GitHub velocity is led by openai/codex; paper attention is clustering around LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling; social attention is tilting toward Ask GN: What do you think about services that involve human intervention in an environment wher…; biggest mover: affaan-m/everything-claude-code (+9). 10 repo signals, 10 paper picks, and 10 community items made today's cut. This fallback brief keeps the page live when the optional Gemma pass times out.
Signal Board
Repositories and papers
Keep the full repo and paper scan above the fold, then read the day as one short brief below.
Top list
Repository Momentum
Fresh GitHub projects worth scanning before the feed turns over.
openai/codex
Lightweight coding agent that runs in your terminal. Updated 1h ago. 81831 stars, +800/7d, created 394d ago.
google-gemini/gemini-cli
An open-source AI agent that brings the power of Gemini directly into your terminal. Updated 1h ago. 103711 stars, +732/7d, created 389d ago.
BerriAI/litellm
Python SDK, Proxy Server (AI Gateway) to call 100+ LLM APIs in OpenAI (or native) format, with cost tracking, guardrails, loadbalancing and logging. [Bedrock, Azure, OpenAI, VertexAI, Coher…
affaan-m/everything-claude-code
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond. Updated 1h ago. 1…
OpenHands/OpenHands
🙌 OpenHands: AI-Driven Development. Updated 1h ago. 73180 stars, +603/7d, created 790d ago.
Significant-Gravitas/AutoGPT
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters. Updated 1h ago. 184171 stars, +311/7…
langfuse/langfuse
🪢 Open source LLM engineering platform: LLM Observability, metrics, evals, prompt management, playground, datasets. Integrates with OpenTelemetry, Langchain, OpenAI SDK, LiteLLM, and more.…
opendataloader-project/opendataloader-pdf
PDF Parser for AI-ready data. Automate PDF accessibility. Open-source. Updated 1h ago. 21085 stars, avg 58.0/day, created 364d ago.
antirez/ds4
DeepSeek 4 Flash local inference engine for Metal. Updated <1h ago. 7184 stars, created 5d ago. Down 9 spots from the previous run.
MemPalace/mempalace
The best-benchmarked open-source AI memory system. And it's free. Updated 7h ago. 51958 stars, +800/7d, created 37d ago.
Top list
Fresh Papers
New research worth bookmarking for a deeper read.
LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling
AutoTTS automates test-time scaling strategy discovery by formulating it as controller synthesis over reasoning trajectories and probe signals, achieving improved accuracy-cost tradeoffs wi…
HyperEyes: Dual-Grained Efficiency-Aware Reinforcement Learning for Parallel Multimodal Search…
HyperEyes is a parallel multimodal search agent that enables concurrent entity searches while optimizing inference efficiency through dual-grained reinforcement learning and a specialized b…
InterLV-Search: Benchmarking Interleaved Multimodal Agentic Search
InterLV-Search benchmark evaluates interleaved language-vision agentic search by repeatedly using textual and visual evidence to condition later search, revealing current systems' limitatio…
Mean Mode Screaming: Mean--Variance Split Residuals for 1000-Layer Diffusion Transformers
Deep diffusion transformers face structural instability at extreme depths due to mean-dominated collapse triggered by mean mode screaming, which is mitigated through mean-variance split res…
A^2RD: Agentic Autoregressive Diffusion for Long Video Consistency
A$^2$RD, an Agentic Auto-Regressive Diffusion architecture, addresses long video synthesis challenges through a closed-loop process with memory tracking, adaptive generation, and hierarchic…
Anisotropic Modality Align
Research addresses the modality gap in multimodal models by proposing an anisotropic geometric correction framework that enables effective unpaired modality alignment through structured rep…
AEM: Adaptive Entropy Modulation for Multi-Turn Agentic Reinforcement Learning
A novel supervision-free credit assignment method for reinforcement learning in language model agents that adapts entropy dynamics at the response level to improve exploration-exploitation…
UniPrefill: Universal Long-Context Prefill Acceleration via Block-wise Dynamic Sparsification
UniPrefill is a prefill acceleration framework that works across various model architectures and integrates seamlessly with vLLM to improve long-context inference efficiency. Surfaced via H…
4DThinker: Thinking with 4D Imagery for Dynamic Spatial Understanding
4DThinker enables vision-language models to perform dynamic spatial reasoning through 4D latent mental imagery, using scalable data generation and novel fine-tuning methods that outperform…
Flow-OPD: On-Policy Distillation for Flow Matching Models
Flow-OPD addresses limitations in Flow Matching text-to-image models through a two-stage alignment approach combining on-policy distillation and manifold anchor regularization, achieving si…
Today in AI
The day in one pass
GitHub velocity is led by openai/codex; paper attention is clustering around LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling; social attention is tilting toward Ask GN: What do you think about services that involve human intervention in an environment wher…; biggest mover: affaan-m/everything-claude-code (+9). 10 repo signals, 10 paper picks, and 10 community items made today's cut.
The quickest scan starts with openai/codex, LLMs Improving LLMs: Agentic Discovery for Test-Time Scaling, and Ask GN: What do you think about services that involve human intervention in an environment wher…, while 10 GitHub-led signals anchor the repo side of the brief.
Use the structured digest below when you need the full wire, raw links, and the longer tail of items that did not make the front narrative.
Archive
Recent issues
Generated from the ranked feed for May 12, 2026 as one single daily issue.
Linked Mentions
No linked mentions yet.