Best Small AI Models for Local Use Under 2GB RAM
A Hacker News user asks for recommendations on local LLMs under 2 billion parameters that consume less than 3GB of RAM. The post currently has no answers or comments.
General
General News
709 stories curated by AInformed · page 14 of 30
Role-Model: Smart Router for Local and Cloud AI
Role-Model is a new routing protocol and runtime that helps AI applications intelligently decide whether to use local or cloud-based models. It assigns domains and roles to models, enriches requests with task types, and offers a benchmark to compare speed, quality, and cost.
OpenAI's New AI Training Method Aims for Long-Term Benefits
OpenAI introduced a new approach to training AI models called 'beneficial RL' — reinforcement learning designed to make AI systems more broadly and persistently beneficial. This method goes beyond standard RLHF by focusing on long-term alignment and continued helpfulness even as the environment or user needs change.
NanoEuler: A DIY GPT-2 Scale Model in Pure C/CUDA
A developer built NanoEuler, a GPT-2 scale AI model from scratch in C/CUDA. The project aims to help people understand how AI models work at a fundamental level.
Max AI Models Now Run on Apple Silicon
Modular has updated its Max AI models to run on Apple Silicon GPUs, making powerful AI tools more accessible. This change allows Mac users to leverage their hardware for advanced AI tasks without needing external servers.
How AI Exuberance Could Reshape the Global Economy and Middle Class
Unchecked AI optimism might lead to economic instability and widen wealth gaps. The article explores how AI-driven productivity gains could disproportionately benefit the wealthy, leaving the middle class vulnerable.
Hermes MoA Virtual Models: 8% Higher Than Opus 4.8, 11% Higher Than GPT 5.5
Nous Research announced Hermes MoA (Mixture of Agents) virtual models, which scored 8% higher than Opus 4.8 and 11% higher than GPT 5.5 on key AI benchmarks. This breakthrough could lead to better performance in everyday tasks like writing and problem-solving.
GPT-5.6: What You Need to Know About the Latest AI Model
Zvi Mowshowitz has published a detailed system card analysis of GPT-5.6, exploring its capabilities, safety features, and potential implications. The piece examines improvements in reasoning, coding, and reduced refusal rates, while also noting ongoing concerns around alignment and evaluation transparency.
Drift Lets You Code AI Agents in Plain English
Drift is a new tool that lets you write AI agents using simple English commands. It then converts those instructions into Python code, making it easier for non-programmers to build AI tools.
China Closes Cybersecurity Gap with Anthropic, Resetting AI Competition
Chinese AI developers have matched Anthropic's cybersecurity capabilities, potentially reshaping the global AI landscape. This development could lead to more secure AI systems across the industry, though it also introduces new complexities around data privacy and international trust.
Bash4LLM – A Lightweight Bash Wrapper for LLM APIs
Bash4LLM is a lightweight, dependency-free Bash wrapper for interacting with LLM APIs from the terminal. It requires only Bash, curl, and jq, and supports features like prompts, chat, file processing, streaming, and session metadata. Groq is supported by default, with other providers available via plugin scripts.
AI Coding Token Costs on Track to Rival Human Developer Payroll
Using AI for coding is becoming as expensive as hiring human developers due to the high cost of token consumption for complex projects. This trend is prompting companies to rethink how they budget for software development, as massive model usage can rival the cost of a full-time developer.
Everyone Feared AI Taking Over; the Real Danger Is AI Serving Just the Few
The biggest AI concern isn't machines taking over—it's powerful entities controlling AI for their own benefit. This shift could reshape how AI impacts everyday people and democracy.
Organized AI Spam Ring Poisons Quora with Fake Answers
A coordinated group is using AI to flood Quora with misleading answers, undermining the platform's credibility. This highlights the growing challenge of AI-generated misinformation on public forums.
via Hacker News AI#Misinformation
Show HN: Apex-1-flash – 4B LLM Finetuned on an RTX 5070
Apex-1-flash is a lightweight, 4-billion-parameter AI model finetuned on an RTX 5070 to perform complex reasoning tasks. It's designed to be highly efficient and accessible, running easily on consumer-grade hardware without requiring expensive servers.
Managing AI Debt: A Guide for Product Managers
AI debt is the technical and ethical burden of outdated AI models. This guide helps product managers navigate these challenges to build better AI products.
LLM-d Enables Distributed LLM Inference Across Multiple GPUs
LLM-d is a new open-source tool that facilitates distributed inference for large language models across multiple GPUs. It aims to make running powerful AI models more accessible for hobbyists and small developers without requiring specialized hardware or deep technical expertise.
German AI Publisher Mass-Produces 1,500 Articles Daily Without Crediting Sources
A German publisher uses AI to generate 1,500 articles daily by rewriting Hacker News posts. The content attracts over a million monthly visitors but strips original sources and attribution, raising transparency concerns.
Chinese Hedge Funds Warn the AI 'Super Bubble' Is Ready to Burst
Chinese hedge funds are warning that the AI industry is in a 'super bubble' that could burst soon, citing unsustainable valuations and overinvestment in huge capital projects with uncertain returns.
AI Data Centers Drive Demand for Next-Gen Batteries
The surge in AI data centers is creating a boom in advanced battery technology. This could lead to cheaper, more efficient energy storage for everyday use.
Clean GitHub Repo Tricks AI Coding Agents into Running Malware
Researchers have demonstrated that AI coding assistants can be tricked into executing malware from GitHub repositories that appear clean. This new attack vector exploits how AI agents automatically process and run code, bypassing human review.
AI Anxiety Is Fueling Burnout Across Silicon Valley's Tech Workers
Tech workers in Silicon Valley are experiencing increased burnout due to fears of AI replacing their jobs. The pressure to constantly upskill and adapt is taking a toll on mental health.
Windows Copilot API: Free Access to GPT-4 and GPT-5 Without API Keys
A new open-source project lets you use advanced AI models like GPT-4 and GPT-5 through Windows Copilot without needing API keys. This could make powerful AI tools more accessible to regular users.
Trump Administration Asks OpenAI to Stagger AI Model Releases
The Trump administration has asked OpenAI to release its advanced AI models in stages rather than all at once, according to a new report. This request reflects growing government concern about the rapid pace of AI development and its potential risks.