Tech News
Delta
Zed editor introduces Delta, a multiplayer coding feature, sparking debate on its utility and AI summaries in code review.
DeepSeek V4 Pro 0813 (on OpenRouter)
The latest DeepSeek Pro model is now available, via API only. I had to link to OpenRouter because DeepSeek don't have any obvious announcement page for their new model.
From assistance to execution: How enterprises put AI to work
OpenAI research reveals how enterprises are adopting agentic AI, using ChatGPT and Codex, and how frontier firms are pulling ahead in AI adoption.
How OneAdvanced deployed over 50 AI agents on UK-sovereign AWS
Learn how OneAdvanced, a UK enterprise software provider, built a UK-sovereign AI platform by self-hosting Llama 4 Maverick and Llama Guard 4 on Amazon SageMaker AI, with a RAG pipeline on pgvector and over 50 agents...
Serve Qwen3.8-2.4T-A95B, a 2.4T-Parameter Model, with Configurable Reasoning on NVIDIA GB300 NVL72
Alibaba released the open weights for Qwen3.8-2.4T-A95B (Qwen3.8-Max), its largest open-weight model, bringing near-frontier capabilities to the open.
GitHub Repos
headroomlabs-ai/headroom
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM.
# Python# agent# aibytedance/deer-flow
An open-source long-horizon SuperAgent harness that researches, codes, and creates.
# Python# agent# agenticascending-llc/jarvis-registry
Connect any AI copilot or autonomous agent to your enterprise tools — through a single, secure MCP/Agent gateway with built-in identity, access control, and full observability.
# Python# agent# agent gatewayResearch Papers
VAKRA: Evaluating Multi-Hop Reasoning Across APIs and Retrieval Under Tool-Use Policies
VAKRA is a new benchmark with over 8,000 executable APIs across 62 domains to test AI agents on multi-hop reasoning and tool use, showing that even the best models struggle with complex tasks and policy constraints.
A Hybrid Nested Harness for Decoupling Structure and Parameters in LLM-Driven Optimization
A hybrid nested search is formalized, in which an outer loop has the LLM propose a structural sketch, with numeric gaps, and an inner numerical optimizer tunes the sketch.
LazyTrain: Limited-resource Allocation toward Zero-waste Yield Optimization in Large Language Model Training
LazyTrain is a new system that helps train large language models on a single GPU by smartly scheduling how data moves between GPU, CPU, and storage. It treats training as a scheduling problem, reducing memory use and speeding up the process.