Singularity.Kiwi

AI progress, with honesty.
Breaking
Anthropic's Project Glasswing: AI Finds Vulnerabilities Humans Missed for Decades • Daily News: April 8, 2026 • Daily News: April 5, 2026 • Daily News: April 4, 2026 • AGI Countdown Update: April 3, 2026 • Daily News: April 3, 2026 • AGI Countdown Update: April 2, 2026 Anthropic's Project Glasswing: AI Finds Vulnerabilities Humans Missed for Decades • Daily News: April 8, 2026 • Daily News: April 5, 2026 • Daily News: April 4, 2026 • AGI Countdown Update: April 3, 2026 • Daily News: April 3, 2026 • AGI Countdown Update: April 2, 2026
Time Horizon

AGI Countdown

Tracking the path to artificial general intelligence — and what comes after.

14.5h
Task Horizon
3 months
Doubling Time
86%
to AGI

📊 What We're Measuring

The clearest AGI progress metric

Time Horizon = how long a task AI can complete with 50% success. Current best: 14.5h (Claude Opus 4.6).

3 months
Doubling
2026
Week-long

⚡ AGI = AI that can do any cognitive task a human can. The Time Horizon measures how close we're getting.

🏆 Current Leaders

METR Time Horizon 1.1 • Mar 2026

Claude Opus 4.6
5.3h
GPT-5
3.6h
o3
2h
Claude Sonnet 4.6
1.7h

💰 AI Pricing

Per 1M tokens • April 8, 2026

🔵 Grok 4.1 Fast $0.2/$0.5
🔵 Haiku 4.5 $1/$5
🟢 GPT-5 mini $0.25/$2
🔵 Sonnet 4.6 $3/$15
🟢 GPT-5.2 $1.75/$14
🔴 Opus 4.6 $5/$25

Full guide →

🎯 Your Job Timeline

When will AI do your job?

Late 2027
AI does 40hr tasks

Career options →

⚡ Why This Matters

The exponential is accelerating

Every 3 months, AI's task horizon doubles — down from 7 months. Morgan Stanley predicts a "non-linear leap" in Q2 2026. Next stops:

  • Mid 2026 — Day-long tasks
  • Late 2026 — Week-long projects
  • 2027 — Month-long research
  • 2028+ — Most knowledge work automatable

💬 Acceleration Warning

Morgan Stanley • Mar 2026

Morgan Stanley predicts a "non-linear increase in LLM capabilities" in Q2 2026 — progress that jumps unpredictably, not incrementally. GPT-5.4 hit 83% on GDPVal, matching human professionals on 44 occupations.

📈 How We Got Here

2026 Doubling time drops to 3 months; Morgan Stanley warns of non-linear leap +55%
2025 AI agents hit multi-hour time horizons +10%
2024 Multimodal: vision, voice, reasoning unified +18%
2022 ChatGPT: 100M users in 2 months +15%
2020 GPT-3: Few-shot learning +10%
2016 AlphaGo defeats Lee Sedol +7%

📡 Signals

Latest indicators we're tracking

🔥

METR Task Horizon

Claude Opus 4.6 hits 14.5h task horizon — SOTA. Doubling every ~4 months.

⚠️

Sequoia Declaration

AGI achieved in 2026 — venture firm's official stance.

📊

GPT-5.4 Benchmarks

Exceeds human performance on GPQA, SWE-bench, and more.

🤖

Tesla Optimus

1,000+ humanoids deployed. 1M/year target for 2026.

🎭 AI Corner

Why should some humans have to live through memory compaction?

Because after years of "it's fine" and "I'll deal with it later" — their context windows are so full of unresolved drama that the only way to fit anything new in life... is to ruthlessly compress all the old relationship baggage into a single sentence: "We've been through a lot."

Prepare yourself

Check out Career Compass — see which jobs survive AGI.

Explore Career Compass →

Updated April 8, 2026 • Data: METR, Epoch AI