Hackernews posts about ARC-AGI

RSS feed for ARC-AGI

01
newA hand-crafted, CPU-only agent clears all 25 public ARC-AGI-3 gameszenodo.org
▲ 2SingarajB 28d discuss
02
newArc AGI 1 400/400 No LLM running on my MacBook
▲ 25CEOs 29d discuss
03
newI Wired a Whole Fly Brain into ARC-AGI-3. It Lost All Season, Then Won One Backwww.kaggle.com
▲ 1MegusAlexandrus 2h 💬 1
04
newShow HN: Cerberus MCP – Flash and sandbox beats Argon 100% vs. 50% on ARC-AGI-2github.com
▲ 1michaelpreid1 2d discuss
05
newDeepSeek v4.1 Flash 1017 Scores on ARC-AGI-1/2arcprize.org
▲ 1matt1 15d discuss
06
newAsk HN: Are AI-written reviews a safety risk? Evidence from ICLR2027 scores
▲ 1fsabau 12h discuss
07
newCodet JSON
▲ 3fifflesticks 26h 💬 2
08
newPerplexity launches AI agents for workwww.perplexity.ai
▲ 61cybermango 16h 💬 54
09
newOpenAI O3 breakthrough high score on ARC-AGI-PUBarcprize.org
▲ 1724maurycy 686d 💬 1755
10
newARC-AGI-3arcprize.org
▲ 498lairv 226d 💬 368
11
newARC-AGI without pretrainingiliao2345.github.io
▲ 351georgehill 612d 💬 121
12
newOpenAI's GPT-6 Astra on ARC-AGI-3arcprize.org
▲ 240vignesh_warar 64d 💬 163
13
newArc-AGI-2 and ARC Prize 2025arcprize.org
▲ 188gkamradt 592d 💬 101
14
newI got the highest score on ARC-AGI again swapping Python for Englishjeremyberman.substack.com
▲ 178freediver 416d 💬 131
15
newARC-AGI Leaderboardarcprize.org
▲ 170rzk 105d 💬 147
16
newSchema Harness Achieves ~99% on Arc‑AGI‑3 Publicschema-harness.github.io
▲ 131jasondavies 113d 💬 83
17
newDay 1 of ARC-AGI-3www.symbolica.ai
▲ 90lairv 225d 💬 76
18
newNvidia AVO scores 100% on the ARC-AGI-3 interactive reasoning benchmarktwitter.com
▲ 71dsrtslnd23 77d 💬 41
19
newO3 "Arc AGI" Postmortemgarymarcus.substack.com
▲ 44signa11 684d 💬 16
20
newEnabling two settings tripled our scores on the ARC-AGI-3 benchmarkopenai.com
▲ 38tedsanders 100d 💬 5
21
newSamsung released a 7M model that achieved 45% on ARC-AGI-1arxiv.org
▲ 34chintler 394d 💬 12
22
newHidden drivers of HRM's performance on ARC-AGIarcprize.org
▲ 31veryluckyxyz 395d 💬 2
23
newGPT-5.5 – No ARC-AGI-3 scores
▲ 25AG25 197d 💬 3
24
newPoetiq shatters ARC-AGI 2 benchmark at half the costpoetiq.ai
▲ 25flavio87 336d 💬 3
25
newTripling an LLM's ARC-AGI-2 score with code evolutionimbue.com
▲ 17danielmewes 252d 💬 2
26
newGrok 4 (Thinking) achieves new SOTA on ARC-AGI-2 (X)twitter.com
▲ 17mellosouls 485d discuss
27
newNvidia AVO achieves 100% in ARC-AGI-3developer.nvidia.com
▲ 13rochansinha 77d 💬 2
28
newARC-AGI 2: Poetiq reaches 75% at less $8 / tasktwitter.com
▲ 13dimtion 318d 💬 1
29
newI beat Grok 4 on ARC-AGI-2 using a CPU-only symbolic engine (18.1% score)github.com
▲ 11kofdai 255d 💬 6
30
newPlayable preview of ARC-AGI-3three.arcprize.org
▲ 11dcre 476d 💬 2
31
newArc‑AGI Without Pretrainingtwitter.com
▲ 10E-Reverance 612d 💬 1
32
newArc AGI-3 100% solved with a skilltwitter.com
▲ 10rochansinha 78d discuss
33
newARC-AGI-3 benchmark is out nowarcprize.org
▲ 9pretext 226d 💬 2
34
newShow HN: ARC-AGI-3 Toolkitdocs.arcprize.org
▲ 9gkamradt 281d 💬 1
35
newARC-AGI-2 human baseline surpassedwww.lesswrong.com
▲ 9hugetim 329d 💬 1
36
newARC-AGI-3 [pdf]arcprize.org
▲ 8sshroot 226d 💬 1
37
newThe Drivers of HRM's Performance on Arc-AGIarcprize.org
▲ 8artninja1988 448d discuss
38
newAgentic systems redraw the Pareto frontier on ARC-AGIpoetiq.ai
▲ 7gkapur 351d 💬 1