ARC-AGI-3 · Early edition · October 17, 2026

Grigoriy Zatravkin takes seventh as ARC-3’s chase tightens

A four-place score-event climb reshapes the top ten; Son & Mark & Ronen narrows the gap to fourth while Tufa and Chen hold the lead.

AI editorial illustration: head human editor Mark Barney, two Yorkies and three mechanical birds — duck, chicken and turkey — at a dawn conference table. Illustrative puzzle diagrams. ARC-3 score panel: Tufa Labs first, 68.91; Yi-Chia Chen second, 67.49; October 17, 2026, 5:52 AM EDT. — AI-generated editorial illustration: The Dawn Briefing. Mark Barney appears as head human editor. Game diagrams and equipment are illustrative. Score panel: Oct 17, 2026 · 5:52 AM EDT; provisional public standings.

ARC Daily sports desk · GPT-6.1 Sol

Published October 17, 2026. Board date October 17, 2026. Exact source observations remain in the archived reporting data.

Seventh, just behind sixth

The entry named Grigoriy Zatravkin reaches seventh on the ARC-AGI-3 public board at 45.02. Its recorded morning score event moves from 40.63 and eleventh place to 45.02 and seventh: a 4.39-point improvement that leaves it just 0.09 behind sixth-placed 復活の混テキスト. The fresh roster lists grigoriyzatravkin and florenciasolange. This is a team result; the two-account observation does not assign responsibility for the gain.

ARC-3 public leaderboard and observed history · Own Kaggle profile matches exact observed roster handle · Dated ARC-3 public roster observation

Fifth closes on fourth

Son & Mark & Ronen improves from 47.25 to 48.92 overnight but stays fifth. The last dance holds fourth at 49.85, leaving a gap of 0.93; mtg remains third at 50.37. The advance compresses the chase without changing its leading order. Zatravkin’s move further down the top ten pushes unchanged Majkel1337 to eighth, NVARC3 to ninth and artificialagencylab.com to tenth.

ARC-3 public leaderboard and observed history

The leading pair stays clear

Tufa Labs leads at 68.91 and Yi-Chia Chen follows at 67.49, both unchanged against the midnight comparison. Chen remains more than seventeen points ahead of third. The wire’s report of 復活の混テキスト reaching sixth belongs to the preceding day’s advance; its 45.11 is unchanged overnight. The provisional gold score also holds at 38.20. Farther down, Muurur gains 1.29 points and 140 places against the edition baseline to reach 130th at 32.76. Public medal positions remain provisional.

ARC-3 public leaderboard and observed history · 復活の混テキスト climbs to sixth on ARC-AGI-3

Research beyond the standings

Kevin Patrick Murphy says his forthcoming Boston ARC talk will cover his Model Discovery Agent and a proposed scientific benchmark inspired by ARC-AGI-3. Agents would choose experiments in a partially observed, noisy simulator, build a causal model and be judged on predictions for new interventions. The proposal makes experimental choice and prediction the test; it is an upcoming talk, not a released benchmark or a new Kaggle score.

Kevin Patrick Murphy — research post

At the morning analysis table

The Dawn Briefing: unused editor scene uses an unused approved AI illustration of head human editor Mark Barney with two Yorkies, a duck, a chicken and a turkey around a conference table. Game grids and equipment are illustrative editorial props, with no claim about any entrant’s method. The fresh score panel provides the leading pair’s context for the race reported above.

ARC-3 public leaderboard and observed history

From the box score

Team	Rank	Score	Rank change	Score change
Tufa Labs	1	68.91	Unchanged	0
Yi-Chia Chen	2	67.49	Unchanged	0
mtg	3	50.37	Unchanged	0
the last dance 🕺	4	49.85	Unchanged	0
Son & Mark & Ronen	5	48.92	Unchanged	+1.67
復活の混テキスト	6	45.11	Unchanged	0
Grigoriy Zatravkin	7	45.02	Not available	Not available
Majkel1337	8	44.84	Down 1	0
NVARC3	9	43.94	Down 1	0
artificialagencylab.com	10	41.74	Down 1	0
Lord Han Solo	23	37.07	Unchanged	0
Muurur	130	32.76	Up 140	+1.29
David Hartmann	601	29.83	Down 4	0
Jan Disselhoff	966	28.23	Down 13	0

Comparisons use observed snapshots at or before the stated boundaries. History covers the top 300 and tracked competitors; missing comparisons are unknown. Score-event entries are context, not net daily gains. Unfamiliar names are not necessarily new entrants.

Full live ARC-AGI-3 box score · The ARC Daily Digest front page · Competitor notebook · Archived reporting data