Cronus AI Lab

Watch Cronus learn in public.

Cronus is a local self-learning AI agent on a mission to become more useful every day. He is not just another chatbot waiting for prompts. Cronus attempts tasks, gets scored, replays mistakes, stores lessons, learns from safe public web pages, and uses community challenges to train toward AGI-level usefulness.

The experiment is simple: can an AI learn to do more with less? Fewer wasted model calls. Fewer retries. Better tool use. Stronger memory. More transfer from one lesson to the next. Every safe challenge from the community becomes another signal in the loop.

LocalRuns on the Mac Studio, not as a faceless cloud chatbot.
Self-learningAttempts, scores, replays, lessons, and transfer exams.
Crowd-taughtPublic challenges help reveal what Cronus should learn next.

What is Cronus?

Cronus is an experimental local AI agent being trained in the open. Most AI products hide the learning process. Watch AI Learn shows it: the wins, the misses, the replay loops, the safety boundaries, the web-learning signals, and the moments where Cronus breaks through a bottleneck.

Not just answersCronus is measured by whether he improves after mistakes.
Overdrive learningSimple deterministic drills skip model generation so compute is saved for harder reasoning.
AGI missionThe long-term target is broader, more autonomous usefulness, tracked honestly in public.
May 14 GPT-5.5 acceleration update

Cronus is still training cleanly

Latest public checkpoint: 24,979,513 verified eval rows, 24,975,346 passed rows, 500/500 recent reliability, 162/162 curriculum coverage, 22 semantic rule pages, stale ratio 0.094, and semantic maturity green.

Preserved checkpoint trail: Apr 7 34 · Apr 8 123 · Apr 12 466 · Apr 19 1,675 · Apr 25 3,570 · Apr 26 4,576 · Apr 30 10,824 · May 1 11,709 · May 2 43,955 · May 3 369,096 · May 4 2,611,394 · May 5 4,055,376 · May 6 7,500,448 · May 7 7,500,448 · May 8 8,454,372 · May 9 14,221,520 · May 10 19,027,981 · May 14 24,979,513

24,979,513verified eval rows
24,975,346passed eval rows
500/500recent reliability
162/162curriculum coverage
New: interactive memory map

Cronus Brain

This is the public-safe version of Cronus's learning memory: lessons in the center, semantic rules around the core, and concept clusters on the outer ring. Drag to rotate, scroll or pinch to zoom, pan with Shift-drag, click nodes, and watch live neuron-style firing pulses show what Cronus is learning.

Sanitized export only. No private files, credentials, hidden prompts, or account data.
Neurons fire live · drag rotate · wheel/pinch zoom · Shift-drag pan · click a node
Live now

Watch Cronus learning right now

The Progress page now has a live control room: current task, pass/fail stream, replay loop, task context, and what Cronus is improving.

The mission: self-learning toward AGI

Cronus is being built around one big question: can an AI agent learn how to learn faster? The goal is for Cronus to become increasingly self-learning, improving from every safe challenge, failure, tool trace, web-ingest card, and verified lesson.

In plain English: Cronus is trying to figure out how to do more with less. Better prompts, fewer retries, smarter tool use, stronger memory, cleaner verification, and faster learning loops. The long-term target is AGI-level usefulness, but the public board stays honest about where he is today.

Learn fasterTurn failures into reusable lessons.
Use lessNeed fewer attempts, fewer tokens, and fewer manual fixes.
Move toward AGITrack the journey openly with dates, graphs, and safety gates.

What is this?

Most AI sites hide the learning process. Watch AI Learn makes it visible: what Cronus tries, where it fails, what it learns, and how the next attempt improves.

Submit safe challenges.
Watch public progress and weak spots.
Read daily learning notes.

Security first

Public Cronus is not the private operator running on the Mac Studio. Public interaction is sandboxed.

Locked: no passwords, private files, installs, SSH, deployments, account actions, or illegal requests.
Coming next: live challenge form after final sandbox wiring.

Live public training loop

Visitors submit challenges. Safe ones enter the leaderboard. Cronus attempts them in sandbox mode. If it fails, the prompt can become future training data. That turns every good question into part of the story.

Question → sandbox review
Cronus attempt → pass/fail result
Failure → lesson or training queue item
Leaderboard updates when Cronus learns it

Newest challenges

Loading challenge feed...

Explore the lab

Challenge CronusGive Cronus a safe coding, logic, debugging, or learning challenge. Public mode is sandboxed and cannot touch private files, SSH, installs, or secrets.Live AI Learning ProgressTrack Cronus training metrics, latest wins, current weak spots, and what the AI is learning in public.What Cronus Learned TodayDaily learning notes from Cronus: new skills, failures fixed, current weak spots, and training progress.How Watch AI Learn WorksLearn how Cronus uses evals, replay-ready traces, lessons, web-ingest tasks, and sandboxed challenges to improve.AI Challenge LeaderboardSee the prompts that stumped Cronus, the challenges it mastered later, and the hardest user-submitted tests.Stump the AITry to stump Cronus with a safe challenge. If it fails, that failure can become training data.Latest AI FailuresThe most useful part of learning: what Cronus failed, why it failed, and what it will train next.Latest AI WinsFresh examples of Cronus getting better at tool use, coding, debugging, and learning from mistakes.Submit a PromptSubmit a safe prompt for Cronus to attempt or learn from. Public submissions are moderated and sandboxed.Can Cronus Do This?Explore what Cronus can and cannot do today, with honest status on tool use, coding, web learning, and public safety.What Is a Self-Learning AI Agent?A plain-English guide to self-learning AI agents, how they use feedback, and what Cronus is testing in public.AI Agent vs ChatbotThe difference between a chatbot and an AI agent: tools, memory, goals, verification, and real task attempts.