Full-text search across 26171 articles. Combine with topic and date filters; results sorted by relevance.
A documentation census of 17 coding agents and harnesses, published on August 17, 2026, found that all 17 transmit code off the local machine in at least one common configuration, and 8 of 17 rows are flagged unresolved …
Graph engineering is the missing fifth layer of AI agent architecture, controlling routing, parallelism, and human approval gates. A developer explains that production failures often stem from under-instrumented layers b…
Pi, a minimal terminal coding harness, achieved a 66.7% pass rate (20 of 30 tasks) in Composio's benchmark using DeepSeek V4 Flash, the best among four newly tested harnesses, and was the cheapest at $0.028 per successfu…
Xaidr, a new open-source Python library, provides in-process runtime security and governance for AI agents, detecting prompt injection, jailbreaks, destructive tool calls, secret leakage, and protocol-level abuse before …
A developer detailed a local-first AI architecture that prioritizes on-device computation with cloud APIs as fallback, citing projects like OpenWork that cut API spend by 87% and improved response times for 73% of reques…
A developer detailed the Scale and Govern phases of the GenAI Cost Lifecycle on AWS, covering batch inference, cross-region routing economics, spot capacity for embeddings, vector storage tiering, and FinOps governance. …
Amazon Web Services (AWS) has instructed its engineers to conserve CPU cycles at all costs as AI workloads, particularly agentic AI systems, strain its cloud infrastructure and cause wait times for CPU server capacity to…
A new public experiment by Prime Intellect ran 153 autonomous runs on the nanoGPT optimizer speedrun across 18 frontier models, finding that Claude Fable 5 and Opus 5 dramatically outperformed others, though no run produ…
Alibaba Group Holding's open-weight Qwen AI models have surpassed 3 billion global downloads in the past six months, making them the world's most-downloaded AI models, according to a Hugging Face report. Qwen has open-so…
Pangram, a $20-per-month web tool, correctly identified AI-generated text in dozens of tests by a technology journalist, including flagging AI sentences inside human essays, but it and Hive Detect failed on several widel…
Investors are concerned about roughly US$70 billion in off-balance-sheet 'phantom liabilities' from AI companies' credit backstops, which could materialize during a downturn. Nvidia's US$500 billion financing partnership…
Cognition's June 2026 rebranding of Windsurf into Devin Desktop with native Agent Client Protocol support, now used by JetBrains, Gemini CLI, GitHub Copilot, and Codex, signals that model choice is becoming composable, s…
Claude Opus 4.7 beat the human world record on the nanoGPT speedrun's optimizer track, reaching the target loss in 2,930 steps versus the human record of 2,990, but a METR study found that AI agents recovered less than 1…
ThoughtDAG, a Show HN project by developer Xia Chen, introduces a graph-based interface that makes LLM context editable, replacing chat with a canvas where edges define the message sequence sent to the model. The app, wh…
Investors are increasingly concerned about roughly $70 billion in off-balance-sheet liabilities tied to AI companies, which could materialize during a downturn, according to Bloomberg. Nvidia Corp.'s $500 billion financi…
A developer fixed a streaming stall in Claude Code by writing a middleware for the claude-code-router proxy that merges SSE deltas into larger chunks. The middleware combines consecutive content_block_delta events only w…
1Password released SCAM, an open-source benchmark that tests AI agents' ability to recognize and report security threats during realistic multi-turn workplace tasks, revealing that a leading AI model opened a phishing li…
A developer's guide outlines the three pillars for building production-ready AI agents: benchmarking, cost optimization, and tooling. It emphasizes task-level evaluation over static benchmarks, using both LLM-as-judge an…
NInfer, a from-scratch C++/CUDA inference engine, achieves up to 1,313.8 aggregate decode tokens per second on a single NVIDIA GeForce RTX 5090 for the Qwen3.6-35B-A3B model at concurrency 8, with the Qwen3.6-27B NVFP4 p…
Qwen released Qwen 3.8 27B, a dense model that runs at approximately 138 tokens/second on an RTX 5090 with the ninfer inference engine, and Z.AI released GLM-5.3, a coding-focused model that achieves Kimi K3-level perfor…