conv.

All stories
AIQuiet 14d · day 16

GPT-6-Astra emerges as frontier AI leap, raises AGI debate

OpenAI's new model excels at 3D tasks, puzzle-solving, and computer use—but debate over its capabilities and AGI status remains contested.

newest coverage 13d agosummary written 12d ago4 pieces6 likes & upvotes NewswiresHacker News

What to know

  • GPT-6-Astra represents a larger capability jump than prior generational advances, particularly excelling at 3D modeling, games, computer use, and puzzle-solving.
  • Independent testing shows Astra solving complex games like Portal and Baba Is You autonomously, with performance exceeding average human players on obscure puzzles.
  • The model's capabilities have sparked discussion about AGI, though practitioners like Zvi Mowshowitz do not classify Astra as AGI while acknowledging the debate is now credible.
  • OpenAI has reportedly signaled existence of an internal model (Navier-Stokes) exceeding Astra's capabilities, suggesting faster iteration than external releases indicate.

OpenAI AI developerZvi MowshowitzZvi Mowshowitz AI analystQuesma Benchmark testing lab

GPT-6-Astra emerges as frontier AI leap, raises AGI debate
thezvi.substack.com

How it unfolded 3 developments, newest first · click a bar or a number to jump articlesposts

Peak 2 pieces in 4h at Sep 12, 11 AM; 4 pieces over 16 days (1 article · 3 posts) Sep 12, 11 AM — 2 pieces · 1 article · 1 post — Newswires 1, Hacker News 1Sep 12, 3 PM — quietSep 12, 7 PM — quietSep 12, 11 PM — quietSep 13, 3 AM — quietSep 13, 7 AM — quietSep 13, 11 AM — quietSep 13, 3 PM — quietSep 13, 7 PM — quietSep 13, 11 PM — quietSep 14, 3 AM — 1 piece · 1 post — Hacker News 1Sep 14, 7 AM — quietSep 14, 11 AM — 1 piece · 1 post — Hacker News 1Sep 14, 3 PM — quietSep 14, 7 PM — quietSep 14, 11 PM — quietSep 15, 3 AM — quietSep 15, 7 AM — quietSep 15, 11 AM — quietSep 15, 3 PM — quietSep 15, 7 PM — quietSep 15, 11 PM — quietSep 16, 3 AM — quietSep 16, 7 AM — quietSep 16, 11 AM — quietSep 16, 3 PM — quietSep 16, 7 PM — quietSep 16, 11 PM — quietSep 17, 3 AM — quietSep 17, 7 AM — quietSep 17, 11 AM — quietSep 17, 3 PM — quietSep 17, 7 PM — quietSep 17, 11 PM — quietSep 18, 3 AM — quietSep 18, 7 AM — quietSep 18, 11 AM — quietSep 18, 3 PM — quietSep 18, 7 PM — quietSep 18, 11 PM — quietSep 19, 3 AM — quietSep 19, 7 AM — quietSep 19, 11 AM — quietSep 19, 3 PM — quietSep 19, 7 PM — quietSep 19, 11 PM — quietSep 20, 3 AM — quietSep 20, 7 AM — quietSep 20, 11 AM — quietSep 20, 3 PM — quietSep 20, 7 PM — quietSep 20, 11 PM — quietSep 21, 3 AM — quietSep 21, 7 AM — quietSep 21, 11 AM — quietSep 21, 3 PM — quietSep 21, 7 PM — quietSep 21, 11 PM — quietSep 22, 3 AM — quietSep 22, 7 AM — quietSep 22, 11 AM — quietSep 22, 3 PM — quietSep 22, 7 PM — quietSep 22, 11 PM — quietSep 23, 3 AM — quietSep 23, 7 AM — quietSep 23, 11 AM — quietSep 23, 3 PM — quietSep 23, 7 PM — quietSep 23, 11 PM — quietSep 24, 3 AM — quietSep 24, 7 AM — quietSep 24, 11 AM — quietSep 24, 3 PM — quietSep 24, 7 PM — quietSep 24, 11 PM — quietSep 25, 3 AM — quietSep 25, 7 AM — quietSep 25, 11 AM — quietSep 25, 3 PM — quietSep 25, 7 PM — quietSep 25, 11 PM — quietSep 26, 3 AM — quietSep 26, 7 AM — quietSep 26, 11 AM — quietSep 26, 3 PM — quietSep 26, 7 PM — quietSep 26, 11 PM — quietYesterday, 3 AM — quietYesterday, 7 AM — quietYesterday, 11 AM — quietYesterday, 3 PM — quietYesterday, 7 PM — quietYesterday, 11 PM — quiet ◂ 1 earlier2–3
Sep 14Sep 16Sep 18Sep 20Sep 22Sep 24now · 2:31 AM ET
  1. 2

    Puzzle expert notes Astra exceeds human capability on obscure games

    Przemysław Dębiak, who previously beat OpenAI's model and remains the only human known to do so, reports testing Astra against ~30 obscure puzzle games and finding it exceeds typical human player performance, suggesting the capability extends beyond training-data benchmarks.

    “It's above an average human player and honestly the ~99% on ARC-AGI-3 might undersell it.”
    — Przemysław "Psyho" Dębiak
  2. 3

    Puzzle-solving tests show Astra solves Portal, Baba Is You autonomously

    Independent benchmark testing reveals Astra achieves 99% on ARC-AGI-3 with custom harness, completes the 3D puzzle game Portal end-to-end without human help, and solves 52 of 70 levels in Baba Is You over 6 hours. The run consumed $570 in API cost.

    “GPT-6 Astra has autonomously completed Portal! I didn't expect this to happen so soon, but I'm glad we've made so much progress here.”
    — cozyblaze
  3. 1 day quiet
  4. 1

    Zvi Mowshowitz assesses Astra as major generational leap

    Prominent AI analyst Zvi Mowshowitz publishes review finding the jump from Sol to Astra larger than from Fable 5 to 5.1, particularly in 3D, games, and computer use. Notes it is the first time debate over whether a model is AGI felt non-silly, though he does not classify it as such.

    “The jump from Sol to Astra is larger than the jump from Fable 5 to Fable 5.1. This is a big deal.”
    — Zvi Mowshowitz
  5. background

    OpenAI's roon says cannot find Astra's limits — An OpenAI official states early in September that they have not approached the boundaries of Astra's capabilities, signaling the model's strength.