Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
NVIDIA and Apple have resolved the hardware bottleneck for on-device AI, with NVIDIA's RTX Spark (Blackwell GPU, Grace CPU, 128GB unified memory) and Apple's M-series chips enabling 4B+ parameter models to run locally. T…
A developer has released Lowfat, a pluggable CLI filter tool that reduced their LLM token usage by 91.8% over two months. The single-binary tool sits between CLI commands and AI agents, stripping verbose output like full…
Microsoft released a tutorial demonstrating how to run a browser-use agent workflow using Microsoft Fara in Google Colab. The tutorial guides users through setting up a mock OpenAI-compatible endpoint to test browser act…
Foyer, an AI startup with about 25 employees, saves $30,000 to $40,000 per month by having each worker pay for personal OpenAI and Anthropic accounts instead of using enterprise plans, CEO Pratyush Rai told Business Insi…
Agentic AI systems, which autonomously plan, execute tasks, and iterate on results, can consume hundreds of thousands to millions of tokens per day, with annual token costs per agent ranging from $1,095 to $3,833. Howeve…
Teams building AI applications are discovering that embedding pipelines for retrieval-augmented generation (RAG) systems are fundamentally a data engineering problem, not a new AI discipline. The process of ingesting, ch…
Google released LiteRT-LM, a new runtime framework that accelerates on-device inference for its Gemma 4 large language models by up to 2.2x using multi-token prediction. The framework, built on the LiteRT platform former…
Zcash dropped 48% after Claude-assisted research uncovered a critical minting vulnerability in the Orchard shielded pool. The flaw could have allowed an attacker to mint funds from the pool, prompting developers to discl…
Researchers published a proof-of-concept AI-driven worm that embeds an open-weight LLM on compromised GPUs to autonomously scan, exploit, and propagate across Linux, Windows, and IoT devices. The worm exposes hosted mode…
Mochallama, a new Java library, enables running llama.cpp inference directly within a Java process using JDK 22's Foreign Function and Memory (FFM) API, eliminating the need for separate daemon processes or native builds…
NVIDIA GPUs handle variable-length inference requests dynamically without recompilation, while TPUs and AWS Trainium require fixed shapes compiled ahead of time, causing crashes or stalls on mismatched inputs. This archi…
Jentic released a command-line tool that scores OpenAPI documents for AI agent legibility, grading them across six dimensions including compliance, security, and discoverability. The Jentic API Scorecard runs locally in …
A developer turned to large language models to extract product data from e-commerce sites with unpredictable HTML, after traditional scraping tools like BeautifulSoup and Scrapy failed due to constantly changing page str…
Tenure has released a new memory system for large language models that claims 100% precision in retrieving relevant context, compared to under 10% for vector search methods. The tool, which operates as a local LLM provid…
A developer built an AI-powered log analysis platform for Spring Boot applications that uses retrieval-augmented generation (RAG) with Ollama models to parse logs, detect exceptions, and explain root causes. The system, …
Anthropic on Thursday called on frontier AI developers to establish a coordinated, verifiable mechanism to pause or slow development if advanced systems begin improving themselves faster than society can manage. The comp…
As AI models approach greater autonomy, developers need practical safety controls—not just policy debates—to manage risk. The most critical feature is a "brake pedal" that can reduce a model's capability, scope, or acces…
A developer replaced a fragile regex-based PDF invoice extraction pipeline with LLM function calling, achieving reliable structured data output after a week of struggling with brittle pattern matching and OCR noise. The …
A Hacker News user challenged the community to identify companies generating real profits from LLM-based products, excluding firms selling infrastructure to AI developers. The user, who described themselves as an "LLM-sk…
Anthropic co-founder Jack Clark warned that artificial intelligence is approaching a point where it could develop without human input, calling for the ability to slow AI progression through government policy and regulati…