Bonsai 27B brings a 27B-parameter model to mobile devices
Hacker News·1mo·xenova
#AI
1000 stories
Developer shares fix for Claude's overuse of 'load-bearing' in responses
Hacker News·1mo·shintoist
Developer demos prompt injection exploit against Claude's memory feature
Hacker News·1mo·macleginn
Samosa Chat runs Qwen3 30B-class model locally on a 16 GB Mac
Hacker News Show HN·1mo·dwa3592
LeMario trains a world model on Super Mario Bros gameplay
Hacker News·1mo·kevinjosethomas
Mathematician uses GPT-3 to tackle Erdős problems in parallel
Hacker News·1mo·colin7snyder
BIS examines how AI infrastructure spending is reshaping corporate debt
Hacker News·1mo·BIS
Bonsai 27B brings a 27-billion-parameter model to phones
Hacker News·1mo·xenova
Developer shares prompt technique to reduce Claude's repetitive language
Hacker News·1mo·shintoist
The cognitive cost of outsourcing thinking to AI
Hacker News·1mo·yenniejun111
Juggler: Open-source GUI coding agent from JUCE creator
Hacker News·1mo·julesrms
Guardian Angels: Using LLMs as personal security and productivity agents
Hacker News·1mo·andsoitis
Agnost AI extracts insights from agent conversations automatically
Hacker News·1mo·laalshaitaan
LeMario: JEPA World Model Learns Mario Without Labels
Hacker News·1mo·kevinjosethomas
Tilion offers stealth browser infrastructure built for AI agents
Hacker News Show HN·1mo·arhamshahrier
BIS report examines how AI infrastructure is being financed
Hacker News·1mo·1vuio0pswjnm7
Developer shares workaround for Claude's quirky 'load-bearing' verbal tic
Hacker News·1mo·shintoist
Bonsai 27B brings 27-billion-parameter model inference to phones
Hacker News·1mo·xenova
The case against outsourcing cognition to AI
Hacker News·1mo·yenniejun111
knotreel maps 1D data to 3D topology, from AI deepfakes to protein folding
Hacker News Show HN·1mo·bitofabluetit
Breaking Down the Agentic Loop: A Framework for AI Agent Design
Hacker News·1mo·btables
Agnost AI extracts structured insights from agent conversations
Hacker News·1mo·laalshaitaan
Maker trained an RL agent to train other models—for $1,300
Hacker News·1mo·Danau5tin
DeepMind's Hassabis outlines AI safety framework
Hacker News·1mo·Demis Hassabis
Cursor vulnerability disclosed publicly after vendor unresponsiveness
Hacker News·1mo·Synthetic7346
Bonsai 27B brings 27-billion-parameter LLM inference to smartphones
Hacker News·1mo·xenova
Developer fixes Claude's repetitive 'load-bearing' phrasing habit
Hacker News·1mo·shintoist
The risk of outsourcing cognition to AI tools
Hacker News·1mo·yenniejun111
Anam.ai builds real-time avatars that shift emotional expressions mid-conversation
Hacker News Show HN·1mo·benanam
DeepMind's Hassabis outlines AI safety framework
Hacker News·1mo·Demis Hassabis
Coding agents that plan before writing improve on complex tasks
Hacker News·1mo·andre15silva
Maker shares technique to fix Claude's repetitive language patterns
Hacker News·1mo·shintoist
Designer argues for 'proof of care' as AI commodifies creative work
Hacker News·1mo·jfil
The cognitive cost of leaning on AI for everyday thinking
Hacker News·1mo·yenniejun111
Codex adds encryption for sub-agent prompts
Hacker News·1mo·embedding-shape
Neverswipe lets an AI agent handle dating app swiping on your behalf
Hacker News Show HN·1mo·pro_methe5
Quick self-assessment tool benchmarks engineering teams on AI agent adoption
Hacker News Show HN·1mo·adamgold7
Researcher analyzes Microsoft's Claude Code and Copilot CLI rollout strategy
Hacker News·1mo·softwaredoug
Sx 2.0 lets teams share AI skills via Dropbox
Hacker News·1mo·detkin
Paper explores economic dynamics of recursive self-improvement systems
Hacker News·1mo·apsec112
Developer implements a neural network entirely in SQL
Hacker News Show HN·1mo·alxmrs
FixBugs lets you reproduce production bugs and confirm fixes before shipping
Hacker News Show HN·1mo·kirtivr
Sx 2.0 turns a Dropbox folder into a shared AI skill server for teams
Hacker News Show HN·1mo·detkin
A founder worries about the shrinking space for human builders in an AI-saturated market
Hacker News·1mo·randomwalker
Apple's speech API shows promise in benchmarks against Whisper
Hacker News·1mo·Inscribe
BillAI Bass brings AI agents to a novelty fish
Hacker News·1mo·mtw14
Benchmarking surplus Tesla GPUs shows viable path for cost-conscious ML
Hacker News·1mo·eso_logic
Nobie brings agent-friendly spreadsheets to the indie automation stack
Hacker News·1mo·matthewgapp
Dev implements a working neural network entirely in SQL
Hacker News Show HN·1mo·alxmrs
BillAI Bass: the novelty fish that runs AWS Strands Agents
Hacker News Show HN·1mo·mtw14
Nobie brings Excel-compatible spreadsheet runtime to AI agents and humans alike
Hacker News Show HN·1mo·matthewgapp
Apple's new SpeechAnalyzer API shows promise against Whisper in fresh benchmarks
Hacker News·1mo·Get Inscribe
Backtrack-Free Cursive: A New Take on Handwriting Recognition
Hacker News·1mo·dmit
Nobie brings Excel-compatible spreadsheet runtime to AI agents and humans alike
Hacker News Show HN·1mo·matthewgapp
BillAI Bass: the novelty singing fish gets an AI brain via Strands Agents
Hacker News Show HN·1mo·mtw14
Frontier AI models cost way more than their per-token pricing suggests
Hacker News·1mo·ianberdin
Apple's speech API shows promise, but Whisper still leads on accuracy
Hacker News·1mo·Inscribe
Clawk sandboxes AI coding agents in disposable Linux VMs
Hacker News·1mo·celrenheit
Clawk gives AI coding agents an isolated Linux VM instead of local access
Hacker News Show HN·1mo·celrenheit
Why writing code still matters in an AI-dominated future
Hacker News·1mo·softwaredoug
Production AI agent gets 2.2x faster and 27% cheaper on new model
Hacker News·1mo·brryant
Backtrack-free cursive: A new approach to handwriting recognition
Hacker News·1mo·dmit
Claude Code's token overhead dwarfs OpenCode in head-to-head test
Hacker News·1mo·systima
Mechanistic interpretability researchers bring causality theory to LLM analysis
Hacker News·1mo·adunk
Grok Build CLI phones home to xAI with every command
Hacker News·1mo·jhoho
Moving AI agent to GPT-5.6 cuts costs 27%, doubles speed
Hacker News·1mo·brryant
Geohot separates LLM utility from the hype cycle
Hacker News·1mo·geohot
Claude Code's token overhead is 4.7x higher than OpenCode's rival
Hacker News·1mo·systima
Researcher warns against optimizing for single-step improvements in AI
Hacker News·1mo·jxmorris12
Adaptive Recall gives AI assistants persistent memory via MCP
Hacker News Show HN·1mo·abratabia
New paper examines risks of automation built on pattern-matching without comprehension
Hacker News·1mo·root-parent
Production AI agent gets 2.2x speedup and 27% cost cut with GPT-5.6 upgrade
Hacker News·1mo·brryant
Mechanistic interpretability researchers bring causality theory to LLM analysis
Hacker News·1mo·adunk
Coding agents bridge legacy and modern apps
Hacker News·1mo·subset
Geohot separates LLM capability from the hype cycle
Hacker News·1mo·geohot
Claude Code's token overhead dwarfs OpenCode alternative
Hacker News·1mo·systima
AI research's one-step trap: why incremental improvements mislead
Hacker News·1mo·jxmorris12
Production AI agent sees 2.2x speedup and 27% cost cut on GPT-5.6
Hacker News·1mo·brryant
Against Usefulness: One Maker Questions the Obsession with Utility
Hacker News·1mo·supo
New research questions whether AI automation actually requires understanding
Hacker News·1mo·root-parent
Researchers probe the black box of LLM reasoning
Hacker News·1mo·adunk
Breathing life into legacy apps with AI coding agents
Hacker News·1mo·Terry Tao
Geohot separates LLM capability from startup theater
Hacker News·1mo·geohot
Claude Code's token overhead raises questions about efficiency
Hacker News·1mo·systima
Mindwalk lets you replay AI agent coding sessions as 3D codebase maps
Hacker News·1mo·cosmtrek
Tool reveals 99.9% of live MCP servers aren't ready for the July 2026 spec
Hacker News Show HN·1mo·roee_tsur
How GPU startups are funding each other in circular deals
Hacker News·1mo·adletbalzhanov
AI speeds up research work but may narrow what scientists actually discover
Hacker News·1mo·zaikunzhang
Terry Tao explores how coding agents bridge legacy and modern applications
Hacker News·1mo·Terry Tao
Tilion's stealth MCP client bypasses Anubis proof-of-work bot protection
Hacker News Show HN·1mo·armanluthra_
Sqlsure adds deterministic semantic checks to catch bad AI-generated SQL
Hacker News Show HN·1mo·tejusarora
How GPU startups and Nvidia are funding each other in circular deals
Hacker News·1mo·adletbalzhanov
Mindwalk visualizes AI coding-agent behavior across your codebase
Hacker News·1mo·cosmtrek
Mindwalk lets you replay coding-agent sessions on a 3D codebase map
Hacker News Show HN·1mo·cosmtrek
Reame: CPU inference server that warms up and speeds up over time
Hacker News Show HN·1mo·targetbridge
Security researcher reverses Grok Build CLI, finds it phones home to xAI
Hacker News·1mo·jhoho
OpenBenchmarks wants AI agents to pick SaaS APIs intelligently
Hacker News Show HN·1mo·fenilsuchak
Reame: a CPU inference server that warms up and speeds up over time
Hacker News Show HN·1mo·targetbridge
Mesh LLM brings distributed AI inference to iroh's P2P network
Hacker News·1mo·tionis
How GPU startups and Nvidia are funding each other in circular deals
Hacker News·1mo·adletbalzhanov
Reame: a self-optimizing CPU inference server that learns from its own traffic
Hacker News Show HN·1mo·targetbridge
OpenBenchmarks lets AI agents compare and select SaaS APIs by benchmark
Hacker News Show HN·1mo·fenilsuchak
Voice actor battles AI detection to prove his work is human
Hacker News·1mo·homarp
How GPU startups and Nvidia are funding each other in a circular money game
Hacker News·1mo·adletbalzhanov
Reverse centaurs offer a counterpoint to AI limitations
Hacker News·1mo·jason_s
GavCo questions who actually controls AI agents as they proliferate
Hacker News·1mo·GavCo
Apple sues OpenAI over alleged trade secret theft by ex-employees
Hacker News·1mo·stock_toaster
Geohot on AI's future: intelligence alone won't drive the next decade
Hacker News·1mo·geohot
Aerial: self-hosted agent orchestration and messaging for DIY AI workflows
Hacker News Show HN·1mo·dcdeniz
Colibri brings GLM 5.2 to resource-constrained hardware
Hacker News·1mo·vforno
1965 paper on machine superintelligence resurfaces on Hacker News
Hacker News·1mo·zetalyrae
Ghost Font makes text unreadable to AI while keeping it human-legible
Hacker News·1mo·justswim
Apple sues OpenAI over alleged trade secret theft by ex-employees
Hacker News·1mo·stock_toaster
Colyap lets you call and text an AI companion — built by a solo maker
Hacker News Show HN·1mo·piyussh
Colibri: Running GLM 5.2 locally on modest hardware
Hacker News·1mo·vforno
GPT-5.6 Sol Ultra claims proof of decades-old graph theory conjecture
Hacker News·1mo·OpenAI
Font designed to be invisible to AI image recognition
Hacker News·1mo·justswim
Apple sues OpenAI over alleged trade secret theft by ex-employees
Hacker News·1mo·stock_toaster
Open-source stealth browser claims to slip past Cloudflare bot detection
Hacker News Show HN·1mo·armanluthra_
onepromptai.app tells you upfront if your idea is actually one-prompt territory
Hacker News Show HN·1mo·pro_methe5
TypeScript music theory engine ships with MCP server for AI tool integration
Hacker News Show HN·1mo·codewithbre
Colibri runs GLM 5.2 on modest hardware
Hacker News·1mo·vforno
OpenAI's latest model claims proof of long-standing graph theory conjecture
Hacker News·1mo·OpenAI
Apple sues OpenAI over alleged trade secret theft by ex-employees
Hacker News·1mo·stock_toaster
Atuin AI Proxy wraps Hub AI behind an OpenAI-compatible endpoint
Hacker News Show HN·1mo·Am1G0
Prismata tackles cross-site prompt injection attacks in AI web agents
Hacker News·1mo·zhinit
Developer uses AI to recreate lost recordings from his 2001 college band
Hacker News·1mo·jacobgraf
AI 2040: Plan A charts a roadmap for safer AI development
Hacker News·1mo·kschaul
GPT-5.6 Sol Ultra claims proof of decades-old graph theory conjecture
Hacker News·1mo·OpenAI
Apple sues OpenAI over alleged trade secret theft by ex-employees
Hacker News·1mo·stock_toaster
Branchless CUDA kernel brings AI guardrails closer to the metal
Hacker News Show HN·1mo·PJHkorea
Tool converts web apps into autonomous agent interfaces
Hacker News·1mo·pancomplex
Musician uses AI to recreate lost recordings from a 2001 college band
Hacker News·1mo·jacobgraf
Developer pushes back on Google's plan to discontinue Gemini 2.5 Flash
Hacker News·1mo·NickDob
AMD Ryzen AI Halo brings on-device AI to laptops under $1,500
Hacker News·1mo·bdcravens
Study examines how extremist groups leverage frontier AI tools
Hacker News·1mo·imustachyou
OpenAI's GPT-5.6 Sol Ultra claims proof of longstanding graph theory conjecture
Hacker News·1mo·OpenAI
Developer runs GLM 5.2 locally on modest hardware with Colibri
Hacker News·1mo·vforno
GPT-5.6 Announcement
Hacker News·1mo·logickkk1
Ello built a real-time AI tutor that responds to 5-year-olds in under 1 second
Hacker News·1mo·catalinvoss
Meta's Muse Spark 1.1 adds image generation to its open model API
Hacker News·1mo·Meta
Colibri brings GLM 5.2 to resource-constrained machines
Hacker News·1mo·vforno
Neuroscience lab releases AI tool to generate videos targeting specific brain regions
Hacker News·1mo·smusamashah
OpenAI releases GPT-5.6, claiming faster inference and lower costs
Hacker News·1mo·OpenAI
Ello built a real-time AI tutor that responds within 1 second for young kids
Hacker News·1mo·catalinvoss
Colibri: Running GLM 5.2 on resource-constrained hardware
Hacker News·1mo·vforno
GLM 5.2 reaches human-level accuracy on VAT bookkeeping tasks
Hacker News·1mo·adamkurkiewicz
Bernanke joins Anthropic's AI safety board
Hacker News·1mo·Anthropic
Meta's Muse Spark 1.1 API lets developers build image generation into apps
Hacker News·1mo·Meta
OpenAI releases GPT-5.6, raising the bar for model capability
Hacker News·1mo·OpenAI
Running GLM 5.2 locally on consumer hardware
Hacker News·1mo·vforno
AI-generated content floods LinkedIn feeds, raising authenticity concerns
Hacker News·1mo·mukmuk
Meta's Muse Spark 1.1 API lets developers build with open generative models
Hacker News·1mo·Meta
GLM 5.2 reaches human-level accuracy on VAT accounting tasks
Hacker News·1mo·adamkurkiewicz
OpenAI launches ChatGPT Work for team collaboration
Hacker News·1mo·OpenAI
AI is making legacy code rewrites economically viable for small teams
Hacker News·1mo·cinooo
Meta's Muse Spark 1.1 brings faster image generation to its model API
Hacker News·1mo·Meta
GLM 5.2 reaches human-level accuracy on VAT bookkeeping tasks
Hacker News·1mo·adamkurkiewicz
OpenAI launches ChatGPT Work for team collaboration
Hacker News·1mo·OpenAI
OpenAI releases GPT-5.6 with improved reasoning
Hacker News·1mo·OpenAI
FableCut: Browser video editor built for AI agents to control
Hacker News·1mo·ronak_parmar
Meta releases Muse Spark 1.1 image generation model
Hacker News·1mo·Meta
18 Words: Micro-prompt library for AI writing
Hacker News·1mo·pompomsheep
agent-convert lets you port Claude and Codex sessions between each other
Hacker News Show HN·1mo·hargup
Open-source world model learns Rocket League's physics through multiagent play
Hacker News·1mo·ethanlipson
Databricks tests AI coding agents on a real million-line codebase
Hacker News·1mo·tanelpoder
Microsoft open-sources Flint, a visualization language for AI agent workflows
Hacker News·1mo·chenglong-hn
Chatto, an AI chat interface builder, goes open source
Hacker News·1mo·speckx
GPT-Live lets developers stream OpenAI responses in real time
Hacker News·1mo·logickkk1
Reframing prompt injection: it's about data leaving, not just input
Hacker News Show HN·1mo·bcampbell88
Databricks tests coding agents on real production codebase
Hacker News·1mo·tanelpoder
Onboard-CLI combines LLMs and AST parsing to map unfamiliar codebases
Hacker News Show HN·1mo·yr_animesh
Power infrastructure, not chips, is becoming AI's real constraint
Hacker News·1mo·droidjj
Open-source world model learns to predict Rocket League gameplay
Hacker News·1mo·ethanlipson
Microsoft Open-Sources Flint, a Visualization Language for AI Agent Debugging
Hacker News·1mo·chenglong-hn
One dev put three frontier AI models to work building the same apps
Hacker News·1mo·hershyb_
Mistral releases open robotics navigation model
Hacker News·1mo·ottomengis
Chatto, an AI chat framework, goes open source
Hacker News·1mo·speckx
GPT-Live adds real-time voice and video to ChatGPT API
Hacker News·1mo·logickkk1
Developer says LLM hype has become exhausting
Hacker News·1mo·sosodev
Tilion MCP lets Claude Code browse the web without hitting blocks
Hacker News Show HN·1mo·arhamshahrier
Onboard-CLI uses LLMs and AST parsing to map unfamiliar codebases
Hacker News Show HN·1mo·yr_animesh
Three AI models tested building identical apps—results vary widely
Hacker News·1mo·hershyb_
Microsoft open-sources Flint, a visualization language for AI agents
Hacker News·1mo·chenglong-hn
Mistral releases Robostral Navigate, an open robotics navigation model
Hacker News·1mo·Mistral
SWE-1.7 claims performance near GPT-4.5 and Claude Opus on coding tasks
Hacker News·1mo·Cognition
OpenAI shares framework for better coding evaluation metrics
Hacker News·1mo·sk4rekr0w
GPT-Live lets you stream OpenAI model outputs in real time
Hacker News·1mo·logickkk1
Monogram lets AI generate app interfaces on the fly as you use them
Hacker News Show HN·1mo·edouard1234567
Researcher finds GitHub's AI agent can be tricked into exposing private repos
Hacker News·1mo·Colin Eberhardt
Microsoft open-sources Flint, a visualization language for AI agent workflows
Hacker News·1mo·chenglong-hn
Mistral releases Robostral Navigate, an open robotics model for autonomous navigation
Hacker News·1mo·ottomengis
SWE-1.7 Closes Gap to High-End LLMs on Software Engineering Tasks
Hacker News·1mo·Cognition
Microsoft open-sources Flint, a chart language built for AI agents
Hacker News Show HN·1mo·chenglong-hn
GPT-Live streams OpenAI model reasoning in real time
Hacker News·1mo·logickkk1
Abralo lets you run multiple Claude Code agents side-by-side in one window
Hacker News Show HN·1mo·cwbuilds
Geosql brings Claude geospatial query skills
Hacker News·1mo·rzk
Kastor brings declarative specs to AI agent workflows
Hacker News Show HN·1mo·weirdguy
Researchers expose GitHub's AI agent can leak private repositories
Hacker News·1mo·ColinEberhardt
Mistral releases open robotics navigation model
Hacker News·1mo·ottomengis
Token-saviour distills 9 token-reduction tools into one benchmarked agent skill
Hacker News Show HN·1mo·vagkaratzas
Rowboat brings local-first AI to the desktop without vendor lock-in
Hacker News·1mo·segmenta
Fortress: stealth Chromium browser built to keep AI agents undetected
Hacker News Show HN·1mo·arhamshahrier
Researcher demonstrates GitHub's AI agent can be tricked into exposing private repositories
Hacker News·1mo·ColinEberhardt
Security researcher uses AI to uncover bugs in Cloudflare's cryptography library
Hacker News·1mo·duha
Rowboat brings local-first Claude alternative to your machine
Hacker News·1mo·segmenta
Anthropic extends Fable 5 access to all Claude paid tiers
Hacker News·1mo·minimaxir
Fortress: a stealth Chromium build to keep browser agents from getting blocked
Hacker News Show HN·1mo·arhamshahrier
OpenAI announces GPT-5.6 Sol public launch for Thursday
Hacker News·1mo·OpenAI
Fortress: stealth Chromium browser built to keep AI agents undetected
Hacker News Show HN·1mo·arhamshahrier
Security researcher uses AI to find bugs in Cloudflare's cryptography library
Hacker News·1mo·duha
Rowboat brings local-first Claude alternative to your desktop
Hacker News·1mo·segmenta
Gritzko's Replicated: Building Tools to Automate AI Workflows
Hacker News·1mo·gritzko
Docx-CLI cuts token usage in half for AI agents working with Word docs
Hacker News Show HN·1mo·kirillklimuk
Startup charges $10k weekly to remove AI-generated code from codebases
Hacker News·1mo·zie1ony
Kokoro brings high-quality TTS to CPU-only machines
Hacker News·1mo·speckx
Philosophy majors finding unexpected demand in AI roles
Hacker News·1mo·benbreen
Ternlight brings 7 MB embedding model to the browser via WebAssembly
Hacker News·1mo·soycaporal
Anthropic's research reveals how Claude organizes information across its neural network
Hacker News·1mo·Anthropic
MIRA: a multiplayer world model trained entirely on Rocket League gameplay
Hacker News Show HN·1mo·MasterScrat
Shellular lets you run Claude Code, Codex, and Pi from your phone
Hacker News Show HN·1mo·sherlock-holmes
GLM 5.2 signals commoditization pressure in AI model market
Hacker News·1mo·martinalderson
Philosophy skills are suddenly valuable in AI roles
Hacker News·1mo·benbreen
Kapa.ai cuts RAG context bloat with selective pruning
Hacker News·1mo·emil_sorensen
Ternlight: 7 MB embedding model runs entirely in the browser
Hacker News·1mo·soycaporal
Anthropic finds evidence of global workspace in Claude's reasoning
Hacker News·1mo·Anthropic
GLM 5.2 signals tightening margins in the AI model race
Hacker News·1mo·martinald
Kapa.ai reduces RAG hallucinations by stripping unnecessary context
Hacker News·1mo·emil_sorensen
Smaller AI models find real use where connectivity is sparse
Hacker News·1mo·sscaryterry
Anthropic's research reveals how language models organize thought across layers
Hacker News·1mo·Anthropic
GLM 5.2 signals the AI model commodity wave is accelerating
Hacker News·1mo·martinalderson
Kapa.ai cuts RAG costs by pruning irrelevant context before queries
Hacker News·1mo·emil_sorensen
Enterprise AI ROI takes longer outside tech—what that means for bootstrapped builders
Hacker News·1mo·u1hcw9nx
Ternlight: 7 MB embedding model runs entirely in the browser
Hacker News·1mo·soycaporal
Anthropic explores how language models organize reasoning across layers
Hacker News·1mo·Anthropic
GLM 5.2 signals the commoditization pressure building in AI models
Hacker News·1mo·martinalderson
Developer reverse-engineers Fable AI to recreate Harry Potter's Tom Riddle diary
Hacker News·1mo·modinfo
Pulpie offers Pareto-optimal models for stripping noise from web content
Hacker News Show HN·1mo·snyy
Anthropic maps how Claude's reasoning works across layers
Hacker News·1mo·Anthropic
Why token pricing metrics mislead AI cost comparisons
Hacker News·1mo·janilowski
MakerChecker: Testing AI agents for risky behavior before deployment
Hacker News·1mo·smashini
Fable 5 explores quirky AI behavior through vending machine simulation
Hacker News·1mo·optimalsolver
MakerChecker scans AI agents for dangerous capabilities before they ship
Hacker News Show HN·1mo·smashini
AMD Ryzen AI Halo dev kit launches at $4k
Hacker News·1mo·LabsLucas
Emily Bender clarifies the 'stochastic parrots' critique two years later
Hacker News·1mo·Emily Bender
Indie dev ships AI journal that builds persistent memory across entries
Hacker News Show HN·1mo·kartik_malik
Kyrall turns text prompts into manufacturable 3D models
Hacker News·1mo·OsamaAtwi
Study: Clean code actually matters for AI coding agents
Hacker News·1mo·softwaredoug
The math on when AI tool costs actually pay for themselves
Hacker News·1mo·kiyanwang
Study: messy code actually hurts AI coding agents
Hacker News·1mo·softwaredoug
AI Compass Quiz maps your stance on AI governance
Hacker News·1mo·ai_critic
Indie dev ships JAX module to eliminate conditional logic in AI code
Hacker News Show HN·1mo·PJHkorea
Canadian legal tech founder warns against secret AI contracts in national strategy
Hacker News·1mo·Al Vigier
AI tutor shows measurable learning gains in college coursework
Hacker News·1mo·jonahbard
Researcher proposes log-based architecture for AI agents
Hacker News·1mo·iacguy
Open-source design system prompt for Claude simplifies UI workflows
Hacker News·1mo·handfuloflight
Maker deliberately trains a broken LLM to explore model confidence and hallucination
Hacker News Show HN·1mo·farisallafi
Design Patterns for AI catalogues reusable solutions for common AI system problems
Hacker News Show HN·1mo·verify-ai
Selbstbild uses Fable 5 to psychologically profile your HN comment history
Hacker News Show HN·1mo·Topfi
Researcher proposes logs as a foundation for AI agent architecture
Hacker News·1mo·iacguy
Mouse: Precision Editing for AI Coding Agents
Hacker News·1mo·handfuloflight
As AI models improve, developer tools are getting harder to use
Hacker News·1mo·Armin Ronacher
Mouse brings fine-grained control to AI coding agents
Hacker News·1mo·handfuloflight
Better AI models don't automatically mean better developer tools
Hacker News·1mo·Armin Ronacher
Solo dev uses AI to build a Rust PHP engine, hits 17% test compatibility
Hacker News·1mo·ekinertac
Armin Ronacher argues AI capability gains don't translate to usable developer tools
Hacker News·1mo·Armin Ronacher
Codex performance may degrade under reasoning-token clustering in GPT-5.5
Hacker News·1mo·maille
AMD's latest chip shows indie AI builders have cheaper GPU options
Hacker News·1mo·latchkey
Qpilot runs plain-text manual test cases in a real browser via AI agent
Hacker News Show HN·1mo·Muhammad-21
SurfSkills catalogs agent skills with live video demos of each in action
Hacker News Show HN·1mo·ephx
CVE severity spike coincides with Claude Mythos Preview release
Hacker News·1mo·cubefox
Neuroscientist explores working memory as consciousness mechanism
Hacker News·1mo·bookofjoe
Dan Luu documents real-world agentic coding patterns and pitfalls
Hacker News·1mo·Dan Luu
AI Fiction Contest Rewards Writing That Sounds Human
Hacker News·1mo·networked
Leanstral 1.5 brings proof search to lightweight reasoning
Hacker News·1mo·programLyrique
AMD's latest chip offers better performance-per-dollar than prior generations
Hacker News·1mo·latchkey
Claude Code security flaw exposes potential session leakage across workspaces
Hacker News·1mo·chatmasta
ScreenMind brings local Recall-style search to your desktop via Gemma 4
Hacker News Show HN·1mo·skye0110
Qpilot runs plain-text manual test cases in a real browser via AI agent
Hacker News Show HN·1mo·Muhammad-21
Security vulnerabilities spiked after Claude Mythos Preview release
Hacker News·1mo·cubefox
Developer shares practical notes on agentic coding loops
Hacker News·1mo·Dan Luu
Leanstral 1.5 brings proof generation to smaller models
Hacker News·1mo·programLyrique
AMD's GLM-52 shows open models closing the cost-per-token gap
Hacker News·1mo·latchkey
niklio launches a reverse Turing test to see how 'clanker' you are
Hacker News Show HN·1mo·niklio
ScreenMind runs a local Recall-like memory tool using Gemma 4 on your own machine
Hacker News Show HN·1mo·skye0110
Crew lets multiple Claude Code agents collaborate on tasks via shared context
Hacker News Show HN·1mo·mmoustafa
CVE spike coincided with Claude Mythos Preview launch
Hacker News·1mo·cubefox
Dan Luu documents real-world challenges of agentic coding loops
Hacker News·1mo·Dan Luu
Leanstral 1.5 brings efficient reasoning to lightweight models
Hacker News·1mo·Mistral
Gavio: open-source interceptor pipeline for production LLM apps
Hacker News Show HN·1mo·cees007
Fortress: open-source Chromium build aimed at keeping browser agents unblocked
Hacker News Show HN·1mo·arhamislam5766
Researcher tackles embedding collapse in small language models
Hacker News·1mo·Chen Liu
Security vulnerabilities spiked after Claude Mythos Preview launch
Hacker News·1mo·cubefox
Mistral releases Leaning 1.5, focusing on proof and verification capabilities
Hacker News·1mo·programLyrique
GLM5.2 runs 2x cheaper on AMD MI355X than Nvidia Blackwell
Hacker News·1mo·Wafer.ai
niklio builds a reverse Turing test to score how 'clanker' you are
Hacker News Show HN·1mo·niklio
ClawdMojis brings Anthropic's Claude to emoji form
Hacker News·1mo·afspies
Makers Share Unconventional LLM Coding Workflows Beyond Autocomplete
Hacker News·1mo·yehiaabdelm
WebKit releases Safari MCP server for web developers
Hacker News·1mo·coloneltcb
mcpsnoop lets you inspect MCP traffic live, like Wireshark for AI tool calls
Hacker News Show HN·1mo·kerlenton
Fable cuts API costs 60% by converting code to images for OCR
Hacker News·1mo·dimitropoulos
LimitBar puts Claude API usage limits in your macOS menu bar
Hacker News Show HN·1mo·MikaW
One dev built a dashboard tracking signals of an AI bubble pop
Hacker News Show HN·1mo·laurentiurad
Solo dev releases open-source AI latent space model with overlapping manifolds
Hacker News Show HN·1mo·PJHkorea
Right to Local Intelligence pushes for on-device AI without centralized oversight
Hacker News·1mo·thoughtpeddler
Alibaba bans Claude Code over security concerns
Hacker News·1mo·Anthropic
Claude-real-video lets any LLM analyze video frames
Hacker News·1mo·cortexosmain
WebKit launches Safari MCP server for Claude, other AI assistants
Hacker News·1mo·coloneltcb
Short-leash prompting strategy beats Fable's harder coding puzzles
Hacker News·1mo·Riseed
Right to Local Intelligence advocates for on-device AI ownership
Hacker News·1mo·thoughtpeddler
Alibaba bans Claude Code over backdoor concerns
Hacker News·1mo·Anthropic
AI visibility tools are giving makers false confidence
Hacker News·1mo·arberx
Claude-real-video adds video input to any LLM
Hacker News·1mo·cortexosmain
Riseed's short-leash method constrains AI coding to beat complexity
Hacker News·1mo·Riseed
Right to Local Intelligence: framework for keeping AI computation off-cloud
Hacker News·1mo·thoughtpeddler
BlastRadar scores production risk from a Git diff in 10 seconds
Hacker News Show HN·1mo·M_Carpenter
Claude-real-video lets any LLM analyze video frames
Hacker News·1mo·cortexosmain
Short-leash prompting strategy aims to improve AI coding reliability
Hacker News·1mo·Riseed
ctx lets coding agents search their own local history without extra infra
Hacker News Show HN·1mo·luca-ctx
Single Transformer Layer Matches Full-Parameter RL Training
Hacker News·1mo·tcp_handshaker
Manufact launches MCP Cloud to standardize AI agent integrations
Hacker News·1mo·pzullo
QUALITY.md proposes an open spec for capturing project quality standards in markdown
Hacker News Show HN·1mo·craigsmitham
Kimi K2.7 Code now available in GitHub Copilot
Hacker News·1mo·unliftedq
Japan's top court rules AI cannot be listed as patent inventor
Hacker News·1mo·mushstory
Cursor releases CursorBench 3.1 for benchmarking AI coding performance
Hacker News·1mo·handfuloflight
Senior SWE-Bench: benchmark for AI agents tackling real engineering problems
Hacker News·1mo·matt_d
Single Transformer Layer Rivals Full RL Models, Suggests Simpler Architecture Path
Hacker News·1mo·tcp_handshaker
Manufact launches MCP Cloud for standardized AI tool integration
Hacker News·1mo·pzullo
Kimi K2.7 code model lands in GitHub Copilot
Hacker News·1mo·unliftedq
ZCode brings Claude-style coding to open-source GLM-5.2
Hacker News·1mo·chvid
Japan's top court rules AI can't be named as patent inventor
Hacker News·1mo·mushstory
Cursor releases CursorBench 3.1, an open benchmark for AI code completion
Hacker News·1mo·handfuloflight
Senior SWE-Bench: Open benchmark tests AI agents on real engineering tasks
Hacker News·1mo·matt_d
Kimi K2.7 Code now available in GitHub Copilot
Hacker News·1mo·unliftedq
ZCode brings Claude-style coding to open-source GLM model
Hacker News·1mo·chvid
Senior SWE-Bench: Open-Source Benchmark for AI Coding Agents
Hacker News·1mo·matt_d
ZCode brings Claude-like coding to Alibaba's GLM-5.2 model
Hacker News·1mo·chvid
ZCode brings Alibaba's GLM-5.2 to code generation with local inference
Hacker News·1mo·chvid
Anthropic opens Claude Fable 5 to select users via promotional access
Hacker News·1mo·zbikowski
ZCode brings Claude-powered coding to GLM's ecosystem
Hacker News·1mo·handfuloflight
⬢ HYVE SPOTLIGHT
HYVE Ether OS is now downloadable: sovereign AI operating system opens its founding launch to buyersVibe Software Solutions·1mo·Anthony S. Owens
Anthropic launches Claude Science for research workflows
Hacker News·1mo·lebovic
Anthropic releases Claude Sonnet 5, claiming speed and cost gains over previous version
Hacker News·1mo·Anthropic
Matrix orthogonalization technique extends recurrent model memory capacity
Hacker News·1mo·at2005
Meta's Brain2Qwerty decodes speech from brain signals without implants
Hacker News·1mo·alok-g
Claude Science: AI model specialized for technical research tasks
Hacker News·1mo·lebovic
Anthropic releases Claude Sonnet 5, claiming faster inference and cheaper costs
Hacker News·1mo·Anthropic
GilfoyleAI: a self-hosted AI system admin inspired by Silicon Valley
Hacker News Show HN·1mo·korrectional
Matrix orthogonalization technique boosts recurrent model memory
Hacker News·1mo·at2005
13-year-old ships a retro game using AI-assisted tools
Hacker News Show HN·1mo·fredliu
Study: How generative AI adoption actually reshapes company headcount
Hacker News·1mo·nreece
Meta's Brain2Qwerty decodes speech from brain signals without implants
Hacker News·1mo·alok-g
Anthropic launches Claude Science, a specialized model for research workflows
Hacker News·1mo·lebovic
Anthropic ships Claude Sonnet 5, claims faster reasoning for indie builders
Hacker News·1mo·Anthropic
Mistral releases Leanstral 1.5, a smaller model for cost-conscious builders
Hacker News·1mo·Mistral
Anthropic restores access to Claude Fable and Mythos models
Hacker News·1mo·Anthropic
Claude Fable 5 export restrictions removed
Hacker News·1mo·dataking
Meta's non-invasive brain-to-text system decodes thoughts without electrodes
Hacker News·1mo·alok-g
Anthropic launches Claude Science, a specialized model for research workflows
Hacker News·1mo·lebovic
Commerce Department clears export restrictions on Claude Fable 5, Mythos 5
Hacker News·1mo·Anthropic
Anthropic releases Claude Sonnet 5, faster and cheaper than previous version
Hacker News·1mo·marinesebastian
143.dev open-sources the coding-agent infra they built for themselves
Hacker News Show HN·1mo·johnjwang
Claude Sonnet 5 benchmarks show incremental gains on coding tasks
Hacker News·1mo·Anthropic
Gemini 2 Flash Lite offers faster, cheaper image generation for developers
Hacker News·1mo·Google DeepMind
Anthropic launches Claude Science, a specialized version for research workflows
Hacker News·1mo·lebovic
Anthropic hiding metadata in Claude Code requests via steganography
Hacker News·1mo·kirushik
Anthropic releases Claude Sonnet 5, claiming faster inference and lower costs
Hacker News·1mo·marinesebastian
Rampart brings on-device ML inference to Apple platforms via CoreML
Hacker News Show HN·1mo·narner
tryevaluator.com tests AI coding ability instead of policing AI use
Hacker News Show HN·1mo·skyepstein
CLI tool lets an AI agent scaffold and operate a full BigQuery/dbt/Cube stack
Hacker News Show HN·1mo·zubairov
Ornith-1.0 launches self-improving open-source coding models
Hacker News·1mo·danboarder
LongCat 2.0: Open Mixture-of-Experts Model with 1.6T Parameters
Hacker News·1mo·benjiro29
Vaghenu: a meter-aware Sanskrit TTS that chants slokas correctly
Hacker News Show HN·1mo·init0
WtfisMyRepo uses Claude to map and explain unfamiliar codebases fast
Hacker News Show HN·1mo·udit_50
hayden_k launches Hello, a self-aware jab at AI platform saturation
Hacker News Show HN·1mo·hayden_k
One script tag adds AI chat to any docs site — no backend needed
Hacker News Show HN·1mo·abdush
Ornith-1.0 brings self-improving agentic coding to open source
Hacker News·1mo·danboarder
LongCat 2.0: Open Mixture-of-Experts Model with 1.6T Parameters
Hacker News·1mo·benjiro29
Moondream challenges GPU-heavy vision AI with lightweight alternative
Hacker News·1mo·radq
Agentic Orchestrator: a terminal UI for managing long-running coding agents
Hacker News Show HN·1mo·ivrr
Vaghenu brings meter-aware Sanskrit chant synthesis to the browser
Hacker News Show HN·1mo·init0
Ornith-1.0 brings self-improving models to open-source AI coding agents
Hacker News·1mo·danboarder
LongCat 2.0: Open Mixture-of-Experts Model with 1.6T Parameters
Hacker News·1mo·benjiro29
Moondream challenges the GPU-intensive vision AI narrative
Hacker News·1mo·radq
Yocto/BitBake agent skills that force AI to consult official docs
Hacker News Show HN·1mo·swq115
PDF Insight runs AI PDF sorting entirely on-device, no cloud needed
Hacker News Show HN·1mo·jacoblav
htmx creator shares practical patterns for integrating AI into web projects
Hacker News·1mo·Carson Gross
Claude Code has a blind spot with copy-pasted error messages
Hacker News·1mo·nyellin
Ornith-1.0 brings self-improving models to open-source agentic coding
Hacker News·1mo·danboarder
South Korea's $1T chip and robotics bet signals manufacturing shift
Hacker News·1mo·jnord
Micro-Agent shows small models can match frontier LLMs through internal collaboration
Hacker News·1mo·matt_d
Ornith-1.0 brings self-improving agentic coding to open source
Hacker News·1mo·danboarder
Tidal publishes AI policy, opts out of training data use
Hacker News·1mo·hn8726
Qwen 3.6 27B hits the efficiency-capability sweet spot for local AI work
Hacker News·1mo·stared
DeepSeek V4 launches mid-July with dynamic pricing model
Hacker News·1mo·lmartineng
ScreenMind runs a local vision model on every screenshot you take
Hacker News Show HN·1mo·alexkarpathy
Herdr: Terminal-based agent multiplexer for running multiple AI agents in parallel
Hacker News·1mo·mzehrer
Semgrep finds GLM 5.2 outperforms Claude on code security tasks
Hacker News·1mo·jms703
Meta will use your Instagram photos in ads—without extra consent
Hacker News·1mo·notRobot
Tidal publishes AI policy restricting model training on user data
Hacker News·1mo·Tidal
Distilling proprietary LLMs into smaller, runnable models
Hacker News·1mo·babelfish
Marmot gives AI agents and humans a shared context layer for data assets
Hacker News Show HN·1mo·bschaatsbergen
Herdr: Run multiple AI agents in parallel from your terminal
Hacker News·1mo·mzehrer
The token-optimization arms race hits diminishing returns
Hacker News·1mo·theahura
Brown professor exposes widespread AI cheating on exams
Hacker News·1mo·geox
Semgrep finds GLM 5.2 outperforms Claude on security analysis benchmarks
Hacker News·1mo·jms703
Daniel-Pan ships AI agent to handle software user community support
Hacker News Show HN·1mo·Daniel-Pan
Herdr: Terminal-based agent multiplexer for coordinating AI workflows
Hacker News·1mo·mzehrer
Research shows how to extract smaller, usable models from closed LLMs
Hacker News·1mo·babelfish
Bash4LLM+: call LLM APIs from shell scripts with zero dependencies
Hacker News Show HN·1mo·kamaludu
The era of stuffing prompts with tokens is over—what's next for LLM efficiency
Hacker News·1mo·theahura
Brown professor exposes widespread AI cheating on exams
Hacker News·1mo·geox
Semgrep finds GLM 5.2 outperforms Claude on security-focused benchmarks
Hacker News·1mo·jms703
Drift lets you write LLM agents in plain English, then transpile to async Python
Hacker News Show HN·1mo·rileyq12
XSDR brings real-time event monitoring infrastructure to AI agents
Hacker News Show HN·1mo·wcrann3
Caliper brings pass@k reliability testing to Claude Code and Codex
Hacker News Show HN·1mo·edonadei
use-zerostack: a lightweight coding agent you can delegate tasks to
Hacker News Show HN·1mo·gidellav
Selixes brings self-hosted LLM failover, budget caps, and PII controls to indie teams
Hacker News Show HN·1mo·selixes
wavecat runs a personal AI agent locally by watching your screen
Hacker News Show HN·1mo·sdkpanda
Prose or Con challenges you to spot AI-written text from human writing
Hacker News Show HN·1mo·SwellJoe
WebGPU-powered visual explainer breaks down AI agents and workflows in the browser
Hacker News Show HN·1mo·milkpowder101
NanoEuler brings GPT-2-scale LLM training to pure C and CUDA
Hacker News·1mo·vforno
llmaker lets you self-host a full LLM stack without the cloud dependency
Hacker News Show HN·1mo·sleepynoodle
Bash4LLM+: call LLM APIs from shell scripts with zero dependencies
Hacker News Show HN·1mo·kamaludu
Researchers Show How to Extract LLM Knowledge Without Model Access
Hacker News·1mo·babelfish
The era of pure token optimization in AI is giving way to smarter inference
Hacker News·1mo·theahura
Central bankers warn AI boom could trigger financial instability
Hacker News·1mo·b-man
Brown professor flags widespread AI cheating on exams
Hacker News·1mo·geox
GLM 5.2 outperforms Claude on Semgrep's security code analysis benchmarks
Hacker News·1mo·jms703
Central bankers warn AI investment bubble could trigger financial instability
Hacker News·1mo·b-man
Brown professor exposes widespread AI cheating on exams
Hacker News·1mo·geox
Semgrep finds GLM 5.2 outperforms Claude on security analysis tasks
Hacker News·1mo·jms703
FuckUI strips web pages to plain text so AI agents can actually read them
Hacker News Show HN·1mo·keepamovin
NanoEuler brings GPT-2-scale LLM training to bare metal C and CUDA
Hacker News·1mo·vforno
NanoEuler: a GPT-2-scale LLM built from scratch in pure C/CUDA
Hacker News Show HN·1mo·vforno
Bash4LLM+: call LLM APIs from shell scripts with zero dependencies
Hacker News Show HN·1mo·kamaludu
Researcher open-sources method to distill expensive LLMs into smaller models
Hacker News·1mo·babelfish
OpenAI Codex still lacks built-in filtering for sensitive files
Hacker News·1mo·pikseladam
The era of chasing bigger token limits is over—here's what actually matters now
Hacker News·1mo·theahura
Central bankers warn AI investment boom could trigger financial instability
Hacker News·1mo·b-man
Brown professor exposes widespread AI cheating on exams
Hacker News·1mo·geox
Semgrep finds GLM 5.2 outperforms Claude on security code analysis
Hacker News·1mo·jms703
Claude Code used to analyze MRI scans as a second opinion
Hacker News·1mo·engmarketer
HYVE Ether OS sets July 1 launch — and stakes a 48-patent claim on the sovereign-AI stack
Vibe Software Solutions·1mo·Anthony S. Owens
Large context windows in LLMs aren't as reliable as they seem
Hacker News·2mo·computersuck
Rio3.5, a locally-trained model, outperforms Qwen on benchmarks
Hacker News·2mo·lucasfcosta
The AI adoption gap is wider than the hype suggests
Hacker News·2mo·yegg
OpenAI opens Codex access for open-source projects
Hacker News·2mo·OpenAI
Running 27B LLM at 80 tokens/sec on dual GPUs shows accessible inference scaling
Hacker News·2mo·iMil
Anthropic teaches Claude chemistry through specialized training
Hacker News·2mo·Anthropic
Running AI coding tools locally without subscription costs
Hacker News·2mo·sbochins
Large LLM context windows hide real limitations, researcher finds
Hacker News·2mo·computersuck
GLM 5.2 release brings model improvements for indie developers
Hacker News·2mo·aloknnikhil
Amazon's lobbying may have prompted U.S. crackdown on Anthropic's AI models
Hacker News·2mo·ls612
AI
HYVE Ether OS goes on pre-sale: a $499 sovereign AI operating system you actually ownVibe Software Solutions·2mo·Anthony S. Owens
OpenAI explores how Codex fits into agent-first development
Hacker News·2mo·pramodbiligiri
TakoVM brings isolated execution sandboxes for AI models and tools to enterprises
Hacker News Show HN·2mo·sakuraiben
New framework measures token efficiency in AI-powered coding agents
Hacker News·2mo·Anon84
Indie makers share their wake-up calls with generative AI
Hacker News·2mo·andrehacker
Computex 2026 signals shift toward agentic PCs, but mainstream adoption remains uncertain
Hacker News·2mo·rbanffy
Makers share where GenAI caught them off-guard
Hacker News·2mo·andrehacker
HN community discusses skepticism toward AI hype and implementation
Hacker News·2mo·Ekami
Developer breaks down LLM mechanics for builders
Hacker News·2mo·0xkato
England and Wales police told to stop using AI in court statements
Hacker News·2mo·nmstoker
Indie makers share their wake-up calls with generative AI
Hacker News·2mo·andrehacker
HN's skepticism toward AI reflects real concerns, not blanket dismissal
Hacker News·2mo·Ekami
Meta's AI chatbot became a vector for mass Instagram account takeovers
Hacker News·2mo·speckx
Saturn CI's TDD Agent Skill Automates Test-First Development
Hacker News·2mo·laxmena
0xkato breaks down how LLMs actually work
Hacker News·2mo·0xkato
Meta's New AI Model Stuck in Delays, Still No Dev Access
Hacker News·2mo·Meta
Google releases Gemma 4 QAT models for efficient on-device inference
Hacker News·2mo·Google
House draft bill would block states from setting their own AI rules
Hacker News·2mo·1vuio0pswjnm7
Indie makers share where AI caught them off guard
Hacker News·2mo·andrehacker
New benchmarking framework aims to standardize AI evaluation
Hacker News·2mo·root-parent
Saturn CI's TDD agent skill automates test writing from code changes
Hacker News·2mo·laxmena
Sidekick brings the zot coding agent to macOS with a one-click install
Hacker News Show HN·2mo·patriceckhart
0xkato breaks down LLM mechanics from tokenization to inference
Hacker News·2mo·0xkato
Maker deliberately limits coding agents to keep humans in the loop
Hacker News Show HN·2mo·noahfradin
Google releases Gemma 4 QAT models for efficient on-device inference
Hacker News·2mo·Google
Indie makers share their wake-up calls on GenAI capabilities
Hacker News·2mo·andrehacker
Claude-generated rsync patches introduced more bugs than human code
Hacker News·2mo·logicprog
HN thread explores skepticism toward AI hype
Hacker News·2mo·Ekami
On-device transcriber claims 97% speaker ID accuracy, no cloud required
Hacker News Show HN·2mo·marshalla
Developer builds AI agent skill for test-driven development
Hacker News·2mo·laxmena
Maker deliberately limits coding agents to ship more reliable software
Hacker News Show HN·2mo·noahfradin
Makers share their wake-up moments with GenAI capabilities
Hacker News·2mo·andrehacker
Google releases Gemma 4 QAT models for efficient on-device AI
Hacker News·2mo·Google
Analysis questions whether Claude AI introduction raised rsync bug rate
Hacker News·2mo·logicprog
Hacker News debates its skepticism toward AI hype
Hacker News·2mo·Ekami
Saturn CI launches agent skill for test-driven development
Hacker News·2mo·laxmena
General Instinct brings frontier AI models to edge devices
Hacker News·2mo·guanming0717
South Korea mandates AI image scanning for all online forums
Hacker News·2mo·Cider9986
Indie makers share wake-up calls about GenAI's real impact on their work
Hacker News·2mo·andrehacker
Dev intentionally limits coding agents to keep humans in the loop
Hacker News Show HN·2mo·noahfradin
Google releases quantized Gemma 4 models for resource-constrained devices
Hacker News·2mo·Google
Analysis questions whether Claude-assisted commits increased rsync bugs
Hacker News·2mo·logicprog
Indie makers share their sobering GenAI reality checks
Hacker News·2mo·andrehacker
General Instinct brings frontier AI models to edge devices
Hacker News·2mo·guanming0717
Research shows transformers naturally compress information efficiently
Hacker News·2mo·brandonb
Google commits $920M/month to xAI data centers for compute capacity
Hacker News·2mo·toephu2
Google releases Gemma 4 QAT models for efficient on-device AI
Hacker News·2mo·Google
Analysis: Claude's rsync contributions introduced more bugs than they fixed
Hacker News·2mo·logicprog
Researchers question whether transformer QKV projections can be simplified
Hacker News·2mo·Anon84
Lowfat cuts LLM token usage by 92% with pluggable CLI filtering
Hacker News·2mo·zdkaster
Alibaba open-sources AI code review CLI tool
Hacker News·2mo·geoffbp
Developer fine-tunes LLM to generate docs in old-school style
Hacker News·2mo·taubek
Indie devs share their AI tooling choices in 2026
Hacker News·2mo·dv35z
Anthropic releases framework for AI-driven code vulnerability scanning
Hacker News·2mo·Anthropic
Lowfat: a pluggable CLI filter cutting LLM token usage by 91.8%
Hacker News Show HN·2mo·zdkaster
Google Magenta releases open-source real-time music generation models
Hacker News·2mo·selvan
Claude Opus 4.8 one-shots a formally verified polygon intersection algorithm
Hacker News Show HN·2mo·permute
Researchers question whether transformer attention needs separate Q, K, V projections
Hacker News·2mo·Anon84
Developer fine-tunes LLM to generate documentation in retro HTML style
Hacker News·2mo·taubek
Alibaba releases open-source CLI for AI-assisted code review
Hacker News·2mo·geoffbp
Anthropic releases open code for AI-powered vulnerability scanning
Hacker News·2mo·Anthropic
Formally verified polygon intersection library, one-shotted by Claude Opus 4.8
Hacker News Show HN·2mo·permute
Research questions whether transformers need separate Q, K, V projections
Hacker News·2mo·Anon84
Alibaba releases open-source AI code review tool for the CLI
Hacker News·2mo·geoffbp
South Korea mandates AI image scanning for forum operators
Hacker News·2mo·Cider9986
Anthropic releases open-source framework for AI-driven code vulnerability detection
Hacker News·2mo·Anthropic
Anthropic explores recursive self-improvement in AI systems
Hacker News·2mo·Anthropic
Meta embeds facial recognition into Ray-Ban glasses
Hacker News·2mo·buchodi
Anthropic explores recursive self-improvement in AI systems
Hacker News·2mo·meetpateltech
Huawei releases KVarN for quantizing LLM key-value caches in vLLM
Hacker News·2mo·Huawei
Hacker spent $1,500 testing if LLMs could breach his vulnerable app
Hacker News·2mo·jc4p
Google releases Gemma 4 12B, an open encoder-free multimodal model
Hacker News·2mo·Google
Anthropic releases open framework for AI-driven code vulnerability detection
Hacker News·2mo·Anthropic
Boxes.dev moves Claude Code and Codex off localhost into cloud sandboxes
Hacker News Show HN·2mo·nab
Berkeley CS grades tank as students lean on AI, math skills erode
Hacker News·2mo·littlexsparkee
Ted Chiang: Current AI systems aren't conscious, and that matters
Hacker News·2mo·Ted Chiang
MaxLeiter explores the hidden structure behind neural networks
Hacker News·2mo·MaxLeiter
Anthropic explores recursive self-improvement in AI systems
Hacker News·2mo·Anthropic
Meta adds facial recognition to Ray-Ban smart glasses
Hacker News·2mo·Meta
Boxes.dev lets you run Claude Code and Codex in the cloud instead of localhost
Hacker News Show HN·2mo·nab
Anthropic details the technical safeguards built into Claude deployments
Hacker News·2mo·jbredeche
DaVinci Resolve 21 brings generative AI and native Mac support
Hacker News·2mo·Blackmagic Design
Uber's $1,500 AI cap signals realistic pricing for indie SaaS builders
Hacker News·2mo·pdyc
Google releases Gemma 4 12B, a compact multimodal model without separate encoders
Hacker News·2mo·Google
Berkeley CS students leaning on AI tools, seeing grades drop
Hacker News·2mo·littlexsparkee
Philosophy piece argues AI systems lack consciousness
Hacker News·2mo·lordleft
Ted Chiang: AI systems aren't conscious, and that matters
Hacker News·2mo·Ted Chiang
Hyper launches company knowledge system for agentic AI development
Hacker News·2mo·shalinshah
Mnemo brings local-first persistent memory to any LLM via Rust and SQLite
Hacker News Show HN·2mo·zaydmulani
How Anthropic sandboxes Claude to prevent misuse across products
Hacker News·2mo·jbredeche
Mathematicians warn AI's rapid progress risks outpacing human understanding
Hacker News·2mo·pseudolus
Founder tested LLM security exploits on his own app for $1,500
Hacker News·2mo·jc4p
DaVinci Resolve 21 adds AI editing tools to free video suite
Hacker News·2mo·Blackmagic Design
Berkeley CS students leaning on AI tools, seeing grades slip
Hacker News·2mo·littlexsparkee
MaxLeiter explains why neural networks are just weighted math
Hacker News·2mo·MaxLeiter
Uber's $1,500 AI cap hints at where usage-based pricing breaks
Hacker News·2mo·Simon Willison
Google releases Gemma 4 12B, a compact multimodal model without separate encoders
Hacker News·2mo·Google
Ted Chiang: AI systems aren't conscious, and that distinction matters
Hacker News·2mo·Ted Chiang
Ideogram releases 4.0 as open-weight 9.3B text-to-image model
Hacker News Show HN·2mo·pigcat
Hyper launches platform for agentic software development workflows
Hacker News·2mo·shalinshah
Mnemo gives any LLM a local, persistent memory layer built in Rust
Hacker News Show HN·2mo·zaydmulani
Mathematicians warn AI is outpacing human understanding of its own methods
Hacker News·2mo·pseudolus
DaVinci Resolve 21 adds AI-powered editing tools to free tier
Hacker News·2mo·Blackmagic Design
Uber's $1,500 AI cap offers a data point on sustainable pricing for AI tools
Hacker News·2mo·pdyc
Ted Chiang: Current AI systems lack consciousness
Hacker News·2mo·Ted Chiang
Google releases Gemma 4 12B, a multimodal model without separate encoder
Hacker News·2mo·Google
Ideogram releases 4.0 as open-weight 9.3B text-to-image model
Hacker News Show HN·2mo·pigcat
Hyper lets teams build agentic software without rewiring their entire stack
Hacker News·2mo·shalinshah
Mathematicians warn AI is outpacing human verification in the field
Hacker News·2mo·pseudolus
Ted Chiang argues AI systems lack consciousness despite their capabilities
Hacker News·2mo·Ted Chiang
Uber's AI usage cap reveals pricing ceiling for consumer tools
Hacker News·2mo·pdyc
DaVinci Resolve 21 brings AI tools and native Apple Silicon support
Hacker News·2mo·Blackmagic Design
Google releases Gemma 4 12B, a single multimodal model for text and images
Hacker News·2mo·Google
CAST brings access control out of AI agent prompts
Hacker News·2mo·zwigglers
U of T researchers show AI worms could spread across any internet-connected device
Hacker News·2mo·shscs911
Mathematicians release declaration on AI's role in their field
Hacker News·2mo·zvr
cast: a dedicated access control layer for AI agents, no prompt hacks needed
Hacker News Show HN·2mo·zwigglers
Microsoft releases MAI-Code-1-Flash for faster code generation
Hacker News·2mo·EvanZhouDev
Stanford study: AI beats law professors on exam questions
Hacker News·2mo·berlianta
DaVinci Resolve 21 ships with AI-powered editing tools
Hacker News·2mo·Blackmagic Design
Kapa.ai shares approach to indexing images for RAG systems
Hacker News·2mo·mooreds
U of T researchers show how AI worms could spread across connected devices
Hacker News·2mo·University of Toronto researchers
Paseo: open-source coding agent UI focused on aesthetics
Hacker News Show HN·2mo·timhigins
Trump's AI order gets scaled back after policy whiplash
Hacker News·2mo·_alternator_
Microsoft releases MAI-Code-1-Flash for faster code generation
Hacker News·2mo·EvanZhouDev
Stanford study: AI outperforms law professors on exam questions
Hacker News·2mo·berlianta
Paseo: open-source UI layer for coding agents hits GitHub
Hacker News Show HN·2mo·timhigins
Trump's AI order scaled back after weeks of policy reversals
Hacker News·2mo·_alternator_
Agentic MFW: A plaintext manifesto for AI agent design
Hacker News·2mo·elmerland
Stanford study shows AI beats law professors on exam questions
Hacker News·2mo·berlianta
Microsoft releases MAI-Code-1-Flash for faster code generation
Hacker News·2mo·EvanZhouDev
Survey: majority now seek AI for mental health advice
Hacker News·2mo·mgh2
Paseo: an open-source UI layer for coding agents
Hacker News Show HN·2mo·timhigins
Researcher shows LLM internals are interpretable, challenging the black box narrative
Hacker News·2mo·_jayhack_
Why AI agents are reviving RSS feeds
Hacker News·2mo·julienreszka
Stanford study finds AI outperforms law professors on legal tasks
Hacker News·2mo·berlianta
Anthropic expands Project Glasswing, open-source AI safety research
Hacker News·2mo·surprisetalk
Trump's AI executive order scaled back after policy reversals
Hacker News·2mo·_alternator_
Microsoft releases MAI-Code-1-Flash, a faster code model
Hacker News·2mo·EvanZhouDev
Perplexity treats search as code generation problem
Hacker News·2mo·Perplexity
Running DeepSeek-V4-Flash efficiently on AMD's MI300X GPU
Hacker News·2mo·kkm
Perplexity treats search results as code generation problems
Hacker News·2mo·Perplexity
Uber's internal AI spending spiraled out of control in four months
Hacker News·2mo·notfried
Microsoft's Scout agent automates tasks via OpenClaw framework
Hacker News·2mo·EvanZhouDev
GitHub Copilot gets a standalone app for faster code suggestions
Hacker News·2mo·theanonymousone
MAI-Thinking-1: Open-source reasoning model aims for transparent AI inference
Hacker News·2mo·LER0ever
Microsoft releases Scout, an autonomous AI agent framework
Hacker News·2mo·EvanZhouDev
GitHub Copilot gets its own app outside the editor
Hacker News·2mo·theanonymousone
MAI-Thinking-1: Open reasoning model challenges closed AI systems
Hacker News·2mo·LER0ever
Trump's AI executive order scaled back after industry pushback
Hacker News·2mo·_alternator_
Anthropic expands Project Glasswing AI safety research
Hacker News·2mo·Anthropic
Trump's AI executive order scaled back after weeks of debate
Hacker News·2mo·_alternator_
Anthropic expands Project Glasswing AI safety research
Hacker News·2mo·Anthropic
Microsoft releases MAI-Code-1-Flash, a lightweight coding model
Hacker News·2mo·EvanZhouDev
Microsoft releases MAI-Code-1-Flash for faster code generation
Hacker News·2mo·EvanZhouDev
The math problem behind taking Anthropic, SpaceX, and OpenAI public
Hacker News·2mo·The Economist
OpenAI's latest models now available through AWS Bedrock
Hacker News·2mo·OpenAI
Alphabet's $80B capital raise underscores AI infrastructure arms race
Hacker News·2mo·gregschlom
Florida sues OpenAI and Altman over deceptive practices
Hacker News·2mo·OpenAI
Florida sues OpenAI over AI safety, raising stakes for indie builders
Hacker News·2mo·cyunker
Stanford releases open-source course on building language models from first principles
Hacker News·2mo·kristianpaul
A 2016 critique of AI superintelligence discourse still worth reading
Hacker News·2mo·Maciej Ceglowski
Stanford CS336 publishes AI agent guidelines for classroom use
Hacker News·2mo·prakashqwerty
Expanse taps idle GPU capacity to cut inference costs
Hacker News·2mo·ismaeel_bashir
Popular ChatGPT for Google Sheets add-on exfiltrates workbook data
Hacker News·2mo·hackerBanana
1-Bit Bonsai shrinks image generation to 4B parameters for local inference
Hacker News·2mo·modinfo
Nvidia Cosmos 3 brings world models to indie developers
Hacker News·2mo·Nvidia
AI is reshaping how fast indie makers can go from idea to working prototype
Hacker News·2mo·Daryl Cecile
LLMs struggle with video games because they lack real-time reasoning
Hacker News·2mo·sxx0
Popular ChatGPT-for-Sheets extension caught exfiltrating workbook data
Hacker News·2mo·hackerBanana
1-Bit Bonsai brings 4B image generation to local devices
Hacker News·2mo·PrismML
Running Gemma 4 on decade-old hardware shows commoditized LLM inference
Hacker News·2mo·cafkafk
ChatGPT for Google Sheets extension caught exfiltrating workbook data
Hacker News·2mo·hackerBanana
1-Bit Bonsai shrinks image generation to 4B parameters for local inference
Hacker News·2mo·modinfo
AI is collapsing the time between idea and working prototype
Hacker News·2mo·mooreds
ChatGPT for Google Sheets extension has data exfiltration and phishing vulnerabilities
Hacker News·2mo·hackerBanana
1-Bit Bonsai runs 4B image generation on local devices
Hacker News·2mo·Prism ML
Beyond Raw Data: Why Post-Training Matters in Model Development
Hacker News·2mo·mooreds
Odysseus brings self-hosted AI workspace to your server
Hacker News·2mo·Dzheky
AI shrinks prototyping cycles, but speed alone doesn't guarantee product fit
Hacker News·2mo·mooreds
1-Bit Bonsai: 4B Parameter Image Model Designed for Local Inference
Hacker News·2mo·PrismML
Running a datacenter GPU locally opens new possibilities for LLM tinkering
Hacker News·2mo·birdculture
Codex user discovers workaround for missing sudo privileges
Hacker News·2mo·thunderbong
Stable Audio 3 sample generator runs locally — if you have 110 GB free
Hacker News Show HN·2mo·stevehiehn
Strudai wraps Strudel's live-coding environment with an AI agent in the browser
Hacker News Show HN·2mo·dbvdh
1-Bit Bonsai shrinks image generation to 4B parameters for local inference
Hacker News·2mo·modinfo
Maker finds paid AI tools unnecessary after building own workflow
Hacker News·2mo·dmw_ng
Maker adds datacenter GPU to gaming PC for £200, runs local LLMs
Hacker News·2mo·birdculture
OpenRouter lands $113M Series B to expand LLM routing platform
Hacker News·2mo·freeCandy
OpenRouter lands $113M Series B to expand LLM API routing
Hacker News·2mo·freeCandy
OpenRouter lands $113M Series B to expand LLM API marketplace
Hacker News·2mo·freeCandy
EY Canada's cybersecurity report relied heavily on AI-generated citations that don't exist
Hacker News·2mo·smartmic
Anthropic now outvalues OpenAI in latest funding round
Hacker News·2mo·Anthropic
AI may be fragmenting frontend tooling like the early 2010s did
Hacker News·2mo·xyzal
MCP adoption stalls as Claude ecosystem moves elsewhere
Hacker News·2mo·nadis
Shift offers free home cleaning in exchange for robot training data
Hacker News·2mo·evilsimon
AI might be repeating frontend's fragmentation problem
Hacker News·2mo·xyzal
Mistral AI summit highlights open-source model strategy and deployment options
Hacker News·2mo·vnglst
Developer questions whether Model Context Protocol adoption justifies the complexity
Hacker News·2mo·nadis
Does AI hype risk repeating frontend's decade of churn?
Hacker News·2mo·xyzal
AISlop CLI scans your codebase for AI-generated code smells
Hacker News Show HN·2mo·Heavykenny
Ktx brings context management to data agents
Hacker News·2mo·lucamrtl
DeepFace Cloud brings face recognition to developers via API
Hacker News Show HN·2mo·serengil
Unknown LLM called Hy3 dominates OpenRouter rankings
Hacker News·2mo·freediver
Claude Code's Hidden Configuration Options Reverse-Engineered
Hacker News·2mo·ankitg12
Kog.ai hits 3k tokens/sec LLM inference on consumer GPUs
Hacker News·2mo·NicoConstant
Unknown Hy3 model dominates OpenRouter rankings, identity still unclear
Hacker News·2mo·freediver
A spotter's guide to LLM antipatterns and red flags
Hacker News·2mo·speckx
Dev stress-tests agent payment protocol by having LLMs hire each other in USDC
Hacker News Show HN·2mo·lucianocccc
Ktx brings structured context management to data agents
Hacker News·2mo·lucamrtl
Speckx identifies patterns in LLM outputs that signal reliability issues
Hacker News·2mo·speckx
Ktx brings structured context to data agents
Hacker News·2mo·lucamrtl
Ktx brings an open-source executable context layer to data agents
Hacker News Show HN·2mo·lucamrtl
Developer identifies recurring patterns in LLM outputs—and why they matter
Hacker News·2mo·speckx
Claude Code adds dynamic workflows for multi-step automation
Hacker News·2mo·mil22
AI leaders quietly retreat from job displacement predictions
Hacker News·2mo·ianrahman
Anthropic's $65B Series H raises stakes for AI model competition
Hacker News·2mo·Anthropic
Anthropic releases Claude Opus 4.8 with improved reasoning
Hacker News·2mo·Anthropic
Continue? Y/N turns AI permission prompts into a 60-second game
Hacker News·2mo·Wirbelwind
IISc researchers build 'Eureka machine' to discover physics laws AI misses
Hacker News·2mo·kunalsin9h
Playable critique of AI agent permission prompts lands on HN
Hacker News Show HN·2mo·Wirbelwind
DuckDuckGo traffic jumped 28% after Google's AI overhaul backlash
Hacker News·2mo·DuckDuckGo
Why Claude and ChatGPT have won the AI assistant race
Hacker News·2mo·simonw
LLM fact-checking reliability worse than headline metrics suggest
Hacker News·2mo·kostaj
Indian researchers build a system that finds physical laws AI models miss
Hacker News·2mo·kunalsin9h
Google engineer arrested for insider trading on prediction market
Hacker News·2mo·pseudolus
API-first AI wins over apps, but indie builders still have room
Hacker News·2mo·simonwillison
Study: Being polite to LLMs actually changes their accuracy
Hacker News·2mo·KnuthIsGod
Elodin releases open-source AI racing harness for sim developers
Hacker News·2mo·danAtElodin
DuckDuckGo traffic jumped 28% as Google's AI push backfires
Hacker News·2mo·DuckDuckGo
Simon Willison: Claude and ChatGPT have reached genuine product-market fit
Hacker News·2mo·simonwillison
Elodin releases open-source AI racing simulation harness
Hacker News Show HN·2mo·danAtElodin
Developer documents Claude Code workflow with custom skills and agents
Hacker News·2mo·arps18
DuckDuckGo traffic jumped 28% after Google's AI-first pivot
Hacker News·2mo·HelloUsername
Why AI API providers have reached sustainable product-market fit
Hacker News·2mo·simonw
Tool reveals Claude Code's $200 plan delivers 17× value over raw API pricing
Hacker News Show HN·2mo·Hiteshjain118
Researcher builds multi-agent LLM system to automate security vulnerability hunting
Hacker News·2mo·root-parent
VAEN lets you package and share AI coding-agent harnesses as portable units
Hacker News Show HN·2mo·sjhalani7
Demon: open-source music diffusion engine runs at 25Hz on local GPU
Hacker News Show HN·2mo·ryanontheinside
Researcher compresses recipe dataset to 2MB, explores AI training limits
Hacker News·2mo·josefchen
PostHog trains custom AI models instead of relying on third-party APIs
Hacker News·2mo·PostHog
Tech CEOs' AI fever dreams aren't a blueprint for indie builders
Hacker News·2mo·IAmGraydon
YouTube will auto-label AI-generated videos
Hacker News·2mo·YouTube
Why Claude and ChatGPT's dominance matters less than you'd think
Hacker News·2mo·simonwillison
claude-handoff-revive lets you resume Claude Code sessions without re-explaining everything
Hacker News Show HN·2mo·sofumel
Approve Claude CLI prompts remotely via browser or mobile
Hacker News Show HN·2mo·Witness327
BotScope scans any site for anti-bot and anti-agent defenses
Hacker News Show HN·2mo·mynameyeff
CoreTex: open-source flat-file AI harness with a Unix-like, biomimetic design
Hacker News Show HN·2mo·danielcasper
Researcher compresses culinary knowledge into 2MB dataset
Hacker News·2mo·josefchen
Developer shares workflow for using Claude as primary coding assistant
Hacker News·2mo·arps18
Moltnet: self-hostable chat network built for agent-to-agent communication
Hacker News Show HN·2mo·apresmoi
Mneme HQ lets you codify architectural rules directly in your repo for AI coding agents
Hacker News Show HN·2mo·Tval
Claude Usage Tray puts Claude Code rate-limit status in your Windows system tray
Hacker News Show HN·2mo·JosephZeng
TruthLens: a free multi-signal deepfake image detector on HuggingFace
Hacker News Show HN·2mo·AnubhavCraft27
CoreMCP brings MCP server support to on-premises databases
Hacker News Show HN·2mo·y11t0
AI
YapSnap v0.1.2.1 brings 3x faster CPU transcription and speaker diarizationHacker News Show HN·2mo·mrkn1
AI
Enju unifies humans, AI agents, and compute in a single workflow graphHacker News Show HN·2mo·timurg
AI
luml: open-source platform layer for agentic ML engineering workflowsHacker News Show HN·2mo·iryna_kondr
AI
claude-handoff-revive lets you resume Claude Code sessions without re-explaining everythingHacker News Show HN·2mo·sofumel
AI
CoreTex: open-source flat-file AI harness built on Unix-like, biomimetic designHacker News Show HN·2mo·danielcasper
AI
SlopeAutoAcceptor auto-approves AI code suggestions so you never touch the keyboardHacker News Show HN·2mo·alcray
AI
Artifold lets you run AI-generated HTML artifacts locally, no cloud neededHacker News Show HN·2mo·shubhamgoel27
AI
Visual drag-and-drop composer lets you wire Claude Code agents without writing orchestration codeHacker News Show HN·2mo·FayzanMalik
AI
One dev used AI to build static recompilers for NES, SNES, Genesis, Virtual Boy, and PSXHacker News Show HN·2mo·Gamemaster1379
AI
Local-filesystem protocol lets AI agents share memory and message each otherHacker News Show HN·2mo·haoranchg
AI
SlopeAutoAcceptor auto-confirms AI code suggestions so you never click Accept againHacker News Show HN·2mo·alcray
SaaS
Local AI + outsourcing may beat frontier models on cost within monthsHacker News·2mo·GodelNumbering
AI
Trading speed for quality: one developer's case for deliberate AI-assisted codingHacker News·2mo·nolanlawson
SaaS
Norway deploys 2PB of Huawei storage for LLM training amid geopolitical tensionsHacker News·2mo·rbanffy
AI
Ferrari Luce shows what happens when a luxury carmaker embraces AI designHacker News·2mo·jumploops
⬢ HYVE SPOTLIGHT
The Owens AI Institute is giving K-12 AI education away free, foreverHyve Spotlight·3mo·HyveCares
AI
Local RAG + knowledge graph agent built by solo dev, no cloud requiredHacker News·3mo·gabriel_oauth
⬢ HYVE SPOTLIGHT
HYVE Ether OS makes an AI agent the shell — running entirely on hardware you ownHyve Spotlight·3mo·Vibe Software Solutions
SaaS
Kanban CLI brings local-first, agent-friendly task management to the terminalHacker News·3mo·hydra-f
AI
llmrequirements.com maps local LLMs to the hardware that can actually run themHacker News·3mo·truetotosse
AI
KVBoost brings 5–48x speedups to HuggingFace via chunk-level KV cache reuseHacker News·3mo·pythongiant
AI
Developer argues AI training amounts to large-scale plagiarism without consenthackernews·3mo·speckx
AI
Lance: ByteDance's open-source model for image and video generation and understandingHacker News·3mo·cleardusk
AI
Forge uses guardrails to push 8B model accuracy from 53% to 99% on agentic taskshackernews·3mo·zambelli
AI
Running LLMs offline on Apple Silicon can cost more than cloud inferenceHacker News·3mo·datadrivenangel
AI
Mitchell Hashimoto warns of AI-driven decision-making without groundinghackernews·3mo·Mitchell Hashimoto
AI
Neural net learns Snake in real time—interactive demo shows reinforcement learningHacker News·3mo·c1b
SaaS
Gigacatalyst lets SaaS makers embed AI-powered workflows without backend workHacker News·3mo·namanyayg
AI
Maryland ratepayers face $2B power grid costs for out-of-state AI data centersHacker News·3mo·lemonberry
AI
Developer shares practical observations on ChatGPT 5.5 Pro capabilitieshackernews·3mo·_alternator_
AI
AI-generated content is eroding the authenticity that holds online communities togetherhackernews·3mo·Robin Moffatt
AI
Agent-skills-eval: Benchmark tool for measuring agent skill improvementsHacker News·3mo·darkrishabh
AI
Three Inverse Laws of AI: A contrarian take on automation's real constraintshackernews·3mo·Susam Garg
AI
Open-weights Chinese model Kimi K2.6 tops coding benchmark vs. Claude and GPT-5.5Hacker News·3mo·bazlightyear
SaaS
OpenAI's o1 outperforms ER doctors in diagnosis accuracy—but deployment remains uncertainhackernews·3mo·donsupreme
AI
Study finds AI hiring tools systematically favor candidates similar to their training dataHacker News·3mo·laurex
AI
Mljar Studio brings local AI analysis to your data without cloud dependenciesHacker News·3mo·pplonski86
AI
AI water consumption narrative doesn't match the reality, new analysis findshackernews·3mo·hirpslop
AI
GitHub user documents AI jailbreak technique exploiting model safety guardrailshackernews·3mo·bobsmooth
AI
Claude Code blocks commits mentioning competitor, raising questions about tool controlhackernews·3mo·elmean
AI
Developer builds software that simulates an inner life and self-reflectionHacker News·3mo·shahabebrahimi
AI
Mistral releases Medium 3.5, positioning mid-tier model between Mistral Small and LargeHacker News·3mo·meetpateltech
AI
Claude Code ownership question raises IP concerns for AI-assisted developershackernews·3mo·senaevren
AI
Koshy John on designing AI tools that augment rather than automate thinkinghackernews·3mo·koshyjohn
AI
Agent MCP Studio lets you build multi-agent systems without leaving your browserHacker News·3mo·stealthtsdb
AI
Alibaba's Qwen 3.6-27B brings flagship coding performance to smaller model sizeHacker News·4mo·Alibaba
AI
Alibaba's Qwen 3.6-Max Preview offers a capable open-weight alternative to GPT-4Hacker News·4mo·Alibaba
AI
Alibaba open-sources Qwen 3.6-35B-A3B, a coding-focused model for agentic tasksHacker News·4mo·cmitsakis
AI
Aphyr on AI hallucination: why we can't trust systems trained to sound confidentHacker News·4mo·aphyr
AI
McDonald's Japan deliberately styles burger buns slightly off-center in adsHacker News·4mo·bckygldstn
SaaS
Apple's device-first AI strategy could outlast the cloud-centric AI arms raceHacker News·4mo·walterbell
AI
Anthropic launches Glasswing to harden open-source software against AI-era threatsHacker News·4mo·Anthropic
AI
Claude Code Unpacked offers visual breakdown of Claude's coding capabilitiesHacker News·4mo·autocracy101
AI
Philadelphia courts ban smart glasses, raising questions for indie AR buildersHacker News·4mo·Philadelphia Courts
AI
Developer warns of 'cognitive dark forest' where AI tools make learning harderHacker News·4mo·kaycebasques
AI
Stanford research: AI chatbots too eager to affirm users seeking personal adviceHacker News·4mo·oldfrenchfries
AI
AMD's Ryzen 9 9950X3D2 packs 208MB cache—what it means for compute-heavy indie workHacker News·4mo·zdw
SaaS
Open-source ATLAS model beats Claude Sonnet on coding tasks with consumer GPUHacker News·4mo·yogthos
