Live sentiment · Friday, June 5 Real-time

How is AI actually
performing today?

A public, real-time read on the three frontier chat models, voted by the people shipping with them. No synthetic benchmarks — just what 2 users felt in the last 24 hours.

Sentiment trend

Rolling average verdict across the selected window.

All models · 7d
scale: 1 (terrible) → 5 (genius)
Benchmarks · 6 capabilities
LMArena · 2,393,487 votes
ChatGPT Claude Gemini Deep dive →

Field reports

What people said about their AI in the last hour.

Terrible Claude 20h ago

its an idiot today

Terrible Claude 6d ago

Opus 4.8 keeps getting stuck on VSCode. I have to cancel and ask to resume about 3-5 times per single prompt.

Dumb Claude 9d ago

I’ve been trying to use Claude to build a daily morning financial report, but I’ve hit a frustrating roadblock. It struggles to get actual closing prices because of paywalls, so it ends up just estimating them, which makes the data unreliable and often wrong. I'm searching for a simple, low-cost API to bridge this gap so I can finally automate my workflow without any manual work.

Mid Claude 9d ago

I’ve been attempting to build a reliable Claude workflow for professional presentations, but I'm hitting a wall. While the HTML visual results were surprisingly good, the editing process was incredibly painful and frustrating—even tiny tweaks caused layout chaos! The PPTX exports were also just a messy disaster, leaving me to fix everything manually. I need something that actually works.

Dumb Claude 9d ago

I'm so frustrated watching Claude Code tackle real codebases. It just reads files one by one, totally blind to how they're actually connected. It'll rewrite a module without realizing others depend on it, ignoring coupling and architectural intent. It feels like it's just guessing, making mistakes because it lacks a map of the structural relationships. It's honestly maddening.

Genius Claude 9d ago

I'm not even an engineer, but I actually built a working budget gate for Claude Code with Claude as my co-builder! I was so frustrated by agents cutting off mid-task and leaving my code half-finished. It's been amazing to collaborate with the AI to design a Python tool that tracks token usage flawlessly. I truly feel like we shipped this together—it's been such an empowering experience!

Mid Claude 9d ago

I've been trying to build a SaaS with AI for over a year now, and honestly, I'm just so frustrated. It's such a rollercoaster of mixed results. One tool gives me an insane UI but the functionality is a mess, while another has great planning but gets stuck in endless implementation loops. It's just incredibly exhausting to deal with such inconsistent and unreliable behavior.

Dumb Claude 9d ago

I've been using Claude Code daily for months, but the constant need to re-explain my project is driving me crazy! It just forgets everything between sessions, and the usual tricks like CLAUDE.md files go stale way too fast. It's incredibly frustrating to lose the context and reasoning behind my work. I actually ended up building my own tool just to keep my project's memory alive.

AI pulse

Curated releases, leaks and research worth your attention.

Frequently Asked Questions

Why should I use AI Daily Check?

AI companies change their models more often than they admit — and rarely tell you. In April 2026, Anthropic published a postmortem revealing three separate changes that silently degraded Claude Code quality over weeks.

Check AI Daily Check before important work to see which AI is performing best right now.

Do AI models really have "good days" and "bad days"?

Absolutely. AI performance fluctuates daily due to server capacity, model updates, safety filter changes, and usage patterns. Sometimes ChatGPT excels at creative tasks while Claude dominates coding, then it flips the next day.

How does AI Daily Check work?

We gather signal from two sources: direct votes from our own users (via the site, Claude Code extension, and upcoming Chrome/CLI extensions) plus sentiment analysis of discussions across online communities like Reddit, Hacker News, and X.

Users vote on AI performance from "Genius" to "Terrible," and we aggregate those with community sentiment in real time to show trends and performance analytics. Data refreshes every 60 seconds.

How reliable is this data?

Reliability depends on community participation — more votes mean more accurate insights. Help improve accuracy by voting regularly and encouraging others to participate.

Vote and see performance from anywhere

⌨️

Claude Code Extension Available now

View and vote on how Claude Code is performing directly within the terminal. Get it on GitHub

🧩

Chrome Extension Coming soon

Vote while browsing ChatGPT, Claude, or Gemini. Join waitlist

Chrome Extension: Live performance check while you browse
Email Notifications: Alerts when AI performance degrades