LIVE — Welcome to the AI Bubble

Everything
AI, One Hub.

Breaking news, model reviews, real-time market data, and community insights — all live, all in one place.

NVDA Live Price
MSFT Live Price
Top Story Source
AI News — Live Feed
Fetching live AI news...
AI News Feed
All OpenAI Google Meta Anthropic Nvidia
Fetching live AI news...
AI Tools & LLMs
All LLMs Image Gen Code Audio Video Agents Productivity
🤖
Claude
LLM — Anthropic
Advanced AI assistant with strong reasoning, coding, and analysis. Praised for nuanced responses and safety focus.
★★★★★
💬
ChatGPT
LLM — OpenAI
Pioneer of consumer AI chat. GPT-4o excels at versatile tasks, multimodal input, and extensive plugin ecosystem.
★★★★½
Gemini
LLM — Google
Google's flagship multimodal model with deep Workspace integration and strong knowledge-heavy performance.
★★★★
🎨
Midjourney
Image Gen
Industry-leading image generation with unmatched aesthetic quality and a strong community.
★★★★★
🦙
Llama 4
LLM — Meta (Open Source)
Meta's open-source powerhouse. Self-host or use via API. Remarkable performance for an openly licensed model.
★★★★½
💻
GitHub Copilot
Code — GitHub / OpenAI
AI pair programmer with deep IDE integration across 40+ languages. Handles completion, chat, and code generation.
★★★★
🎙️
ElevenLabs
Audio — Voice AI
Industry-leading voice synthesis and cloning. Generates natural speech in dozens of languages with fine-grained emotional control.
★★★★★
🎬
Runway
Video Gen — Runway
Generative video and editing suite. Text-to-video and video-to-video tools used widely across film, ads, and social content.
★★★★½
🔍
Perplexity
Agents — Research AI
AI-powered search and research agent. Answers with live web results and cited sources, with deep-research modes for longer reports.
★★★★½
📝
Notion AI
Productivity — Notion
AI writing, summarization, and Q&A built directly into Notion workspaces. Turns scattered notes and docs into searchable knowledge.
★★★★
🐋
DeepSeek
LLM — DeepSeek (Open Source)
Chinese open-weight lab whose V4 models approach Western frontier performance at a fraction of the API cost, with a 1M-token context window.
★★★★½
🌬️
Mistral
LLM — Mistral AI (Open Source)
European open-weight champion from Paris, balancing strong multilingual performance with the data-sovereignty story enterprises and governments want.
★★★★
🌲
Black Forest Labs
Image Gen — FLUX (Open Source)
Creators of the FLUX image model family — some of the highest-quality open-weight image generators, iterating faster than most closed competitors.
★★★★½
⌨️
Cursor
Code — AI Editor
AI-native code editor built around frontier models with deep codebase awareness. One of the fastest-growing developer tools ever shipped.
★★★★★
🎵
Suno
Audio — Music Gen
Generates complete songs — vocals, instruments, and lyrics — from a text prompt. The fastest path from idea to finished track.
★★★★½
🎭
HeyGen
Video — AI Avatars
Turns a script and a face into a polished talking-head video. Heavily used for personalized sales outreach and localized marketing at scale.
★★★★
🤲
Manus
Agents — Autonomous AI
General-purpose autonomous agent that plans and executes full multi-step tasks — browsing, coding, and research — from a single prompt.
★★★★
📊
Gamma
Productivity — AI Decks
Turns rough notes or a single prompt into a polished, presentation-ready deck or document. One of the most capital-efficient AI products ever built.
★★★★½
Education — Learn AI & LLMs

A curated roadmap of the best courses for understanding and building with AI — from zero-code introductions to hands-on LLM engineering, spot-checked for 2026 relevance.

All Free Paid / Certificate
🧪
ML Crash Course
Google · Self-Paced
Google's foundational machine learning course, refreshed for 2026 with expanded LLM and agentic AI modules. Interactive exercises run in Colab.
🤗
Agents Course
Hugging Face · Certificate
Learn to build autonomous AI agents that plan tasks, call tools, and execute multi-step workflows — arguably 2026's most in-demand skill.
💬
Prompt Engineering for Developers
DeepLearning.AI
A short, hands-on intro from Andrew Ng and Isa Fulford. Build a custom chatbot while learning core prompting techniques with a real LLM API.
📘
AI for Beginners
Microsoft · 24 Lessons
An open-source curriculum covering neural networks, NLP, computer vision, and ethics — published on GitHub with code notebooks for every lesson.
🎓
CS50's Intro to AI with Python
Harvard
A rigorous, project-based dive into search algorithms, logic, probability, machine learning, and NLP. Lectures on YouTube, labs in Python.
🧑‍💻
Practical Deep Learning for Coders
fast.ai
A code-first deep learning course for people who already write software. Build real models from lesson one — no PhD required.
🟩
Deep Learning Institute
NVIDIA · Self-Paced
NVIDIA's free training catalog covering generative AI fundamentals, accelerated computing, and applied deep learning — straight from the hardware maker.
📐
Machine Learning Specialization
Stanford · Free to Audit
The legendary Andrew Ng course, reborn. Covers supervised learning, neural networks, and unsupervised learning — the foundation everything else builds on.
🏗️
Generative AI with LLMs
DeepLearning.AI × AWS
The full LLM lifecycle — pretraining, fine-tuning, RLHF, and deployment — with hands-on labs on real cloud infrastructure. ~16 hours.
🔧
Generative AI Engineering
IBM · Professional Certificate
Build real LLM-powered apps using LangChain, RAG pipelines, and open models like Llama — a full professional certificate track.
☁️
Generative AI Learning Path
Google Cloud · Vertex AI
Vendor-aligned training for building on Gemini and Vertex AI — ideal if your stack already runs on Google Cloud.
🟧
Generative AI Essentials
AWS · New for 2026
A practical on-ramp to the AWS generative AI stack — Bedrock, foundation models, and responsible AI practices for builders.
🔷
Azure AI Engineer Associate
Microsoft Certification
Covers Azure AI Foundry, fine-tuning with Azure OpenAI Service, and end-to-end MLOps for generative AI — a recognized industry credential.
Homebrew — Run AI At Home

Self-hosting an LLM means total privacy, zero subscription fees, and offline access — at the cost of needing real hardware. Here's the current toolchain for running your own models, plus what to expect from your GPU.

🦙
Ollama
CLI · API Server
The fastest path to a local, OpenAI-compatible API. One command — ollama run llama3.2 — and you're chatting with a model in seconds. The backbone of the local LLM ecosystem.
🖥️
LM Studio
Desktop GUI
The most polished interface for discovering, downloading, and chatting with local models. Browse a built-in catalog and click to run — zero command line needed.
🌐
Open WebUI
Self-Hosted Web App
A ChatGPT-style browser interface that layers on top of Ollama. Multi-user accounts, chat history, and model switching — your own private ChatGPT.
📚
AnythingLLM
RAG / Document Chat
Point it at a folder of PDFs and notes — AnythingLLM handles chunking, embeddings, and retrieval automatically. A fully local NotebookLM, built on Ollama or LM Studio.
⚙️
llama.cpp
Inference Engine
The lean C++ engine quietly powering most of the local LLM ecosystem. Maximum efficiency and control for users who want to tune every parameter.
🚀
vLLM
Production Serving
A high-throughput serving engine built for running models at scale with multiple concurrent users — the same tech many inference providers run under the hood.
🪄
Jan
Offline Desktop Assistant
An open-source, ChatGPT-style assistant that runs entirely offline — clean UI, built-in model hub, and no account required.
🪟
GPT4All
Desktop App
A simple, polished app for running local models with minimal setup — especially friendly for Windows users taking their first steps into local AI.
Hardware Guide — Build or Buy

Two paths to running AI at home: buy something that works out of the box, or build a machine around a GPU. Neither is "wrong" — it depends on whether you value simplicity and silence, or raw throughput and upgrade flexibility.

Plug & Play
🖥️
Mac Mini M4 Pro
64GB Unified Memory
Whisper-quiet at ~40W, yet benchmarks show it outperforming a dual RTX 3090 rig (700W) on 32B model inference. The default recommendation for most home setups.
🎛️
Mac Studio M4 Max
Up to 128GB Unified Memory
When 64GB isn't enough. Same efficient architecture, with room to load 70B-class models without aggressive quantization.
🔲
NVIDIA DGX Spark
128GB Coherent Unified Memory
NVIDIA's compact AI supercomputer for the desk. CUDA-native and purpose-built for running and fine-tuning large models locally.
📦
Budget AMD Mini PC
32GB DDR5 · iGPU
Proof that local AI doesn't require a big budget. Runs 7B models at roughly 18-25 tokens/sec via integrated GPU acceleration.
Budget Tier — Build
🟦
RTX 3060 12GB
12GB GDDR6 · ~$200-250
The clear budget champion. Runs every 7-8B model at Q4/Q5 and most 13-14B models at Q4 — Qwen3 14B hits 9-12 tok/s. Works out of the box with Ollama and llama.cpp.
🔵
Intel Arc B580
12GB GDDR6 · New Stock
Newer than the others here, but currently the best price-per-gigabyte on the market for entry-level 7B inference — worth checking if 3060s are scarce near you.
⚠️
Tesla P40 24GB
24GB GDDR5 · ~$150-200
The cheapest 24GB you'll find — but an ex-datacenter card from 2016. No display output, needs an aftermarket fan, and modern driver support is shaky. Only worth it if "won't fit" is your problem, not "too slow."
Performance Tier — GPUs
🟩
RTX 5090
32GB GDDR7 · 1,792 GB/s
The current speed king — around 185 tok/s on 8B models. Needs an 850-1000W PSU and serious case airflow. Street prices remain elevated due to ongoing DRAM shortages.
🟢
RTX 4090
24GB GDDR6X · ~1,008 GB/s
Last generation's flagship, still excellent. Roughly 25-35% slower than the 5090, with easier availability. Needs a 750W+ PSU.
♻️
RTX 3090 (Used)
24GB GDDR6X · ~936 GB/s
The consensus value pick once you're ready to move up. Same 24GB VRAM as a 4090 for roughly a third of the price, at about 15-20% slower token generation.
🔗
Dual-GPU Setup
2× RTX 3090 = 48GB VRAM
Split 70B models across two cards via llama.cpp tensor parallelism. Two 5090s (64GB combined) have reportedly beaten an H100 on Llama 3.3 70B.
Loading live prices...
AI Market Tracker
Combined Market Cap
Loading 15 AI-relevant companies...
AI Bubble Sentiment
Calculating from live price action...
Fetching live market data...
Company Profile
Loading company data...
The Blog
JM
Jordan Mills
Trending
Opinion

Why I Stopped Using ChatGPT and Never Looked Back

After 18 months of daily use, I made the switch to a different AI stack. Here's what I discovered about my own workflow.

SR
Sam Reeves
Deep Dive

I Built a Full SaaS App Using Only AI Agents — Here's What Broke

An eight-week experiment using autonomous coding agents to ship a production product. The wins were huge. The failures were instructive.

AK
Asha Kapoor
Ethics & Safety

The Alignment Problem Isn't What You Think It Is

Most public discourse around AI alignment misses the core issue. Here's a clearer framing of what we need to get right.

Community Forum
Model Discussion
Compare, benchmark, and debate AI models
GPT-5 vs Claude — which handles multi-step reasoning better?
by technerd_42 · 23 mins ago
84
2.1K
Llama 4 local setup guide — M3 Max benchmarks inside
by localrunner · 2 hrs ago
41
890
Projects & Builds
Show off what you're building with AI
[Show] I built an AI-powered code reviewer that runs locally
by devmagician · 1 hr ago
102
3.4K
Ethics & Safety
AI safety, alignment, policy, and societal impact
Should AI companies be legally liable for model outputs?
by policywatch · 4 hrs ago
193
5.8K
Merch Store
AllApparelAccessoriesStickersLimited Drops
🖤
Neural Net Tee
Premium cotton · 4 colors
🧢Limited
AI Bubble Snapback
Structured fit · Black/Neon
🎒
Techpack Backpack
30L · Padded laptop sleeve
Prompt Engineer Mug
15oz ceramic · Dishwasher safe
📦
Sticker Pack Vol. 1
12 holographic stickers
👕New
Alignment Hoodie
400gsm fleece · Oversized fit
LIVE
NVDA Loading... Connecting to live data... NVDA Loading... Connecting to live data...