OpenAI has lowered the context window size of the Codex model from the existing 372k to 272k.
Analysis of LLVM commits predicts the architectural specifications of AMD's next-generation AI accelerator MI455X (GFX1250).
An open-source Claude Code plugin that automatically handles everything from analyzing GitHub issues to generating PRs has been released.
A high-performance, multi-model speech recognition (ASR) library built on ggml has been released.
A new tool called 'Mindwalk' visualizes a coding agent's task context as a 3D night map, revealing exploration and modification processes that logs alone couldn't show.
Moonshot AI has unveiled Kimi K3, a large-scale open-weight model with 2.8 trillion parameters.
A new KV cache technique that stores verified knowledge as KV states and restores it without recomputation was announced.
This is a summary of a survey paper that organizes 25 deep learning methodologies applied to scRNA-seq analysis into 6 categories.
ik_llama.cpp has added support for the openPangu-2.0-Flash model along with a GGUF format update.
The promotion offering an extra 50% weekly usage for Claude Code has been extended until August 19, 2026.
It implements a block-diagonal attention kernel that is up to 3.5x faster than Flash Attention v2 on NVIDIA Blackwell GPUs.
Google DeepMind and Isomorphic Labs have announced an AI-driven biosecurity and disease response strategy.
Wes McKinney, the creator of Pandas and Apache Arrow, analyzes the changes in software engineering and the limitations of LLMs in the AI era.
Anthropic shipped an undisclosed AFK timeout feature to Claude Code and disabled it two days later.
Anthropic announced that Claude's top-performing model Fable 5 will be permanently included in all paid subscription plans starting July 20.
LangChain has released a hosted service that automatically extracts structured data from unstructured documents.
Instead of using hosted APIs, Netflix built its own LLM serving infrastructure to directly operate the entire stack from model deployment to inference.
At least 105 founders from Y Combinator have joined OpenAI (70) and Anthropic (35).
Bulgarian social enterprise Stray Sheep is expanding educational content for the deaf by adding AI voice to sign language using ElevenLabs' text-to-speech technology.
As AI makes code production cheap, the criterion for judging feature requests has shifted from implementation cost to ownership cost.
Google has released 13 hands-on demos for learning the Gemini Enterprise Agent Platform.
NVIDIA and HuggingFace have unveiled a NeMo Automodel integration for fine-tuning diffusion models.
Claude Sonnet 5 introduces a new tokenizer that generates 1.4x more tokens for the same text, so despite unchanged list prices, actual costs will rise by more than 30% starting in September.
NVIDIA Vera Rubin maximizes Intelligence per Dollar for Post-Training workloads, leading the economics of the Agentic AI era.
AI Briefing