VQV Signal

ROBOTS & HARDWARE WATCH TECHNICAL

GPT-6.2 Astra reportedly tops a real-robot benchmark; ARC-AGI-3 score needs a caveat

According to a post on X by @thsottiaux, OpenAI's Thibault Sottiaux called the new GPT-6.2 models clearly capable but polarizing, with rough edges still to fix. The source also says RoboCurve reports that GPT-6.2 Astra tops its RoboDojo-RC Tier 1 real-robot benchmark, at about 70 times the cost of the robot model Dream-1. A separate claim says Artificial Analysis ranked it first on a benchmark, but the source is cut off before naming it. The source is a digest of these claims, so they have not been independently checked here.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-18T06:25:00+00:00 Detected 2026-11-18T20:03:28+00:00
View original source

According to a post on X by @thsottiaux, OpenAI's Thibault Sottiaux called the new GPT-6.2 models clearly capable but polarizing, with rough edges still to fix. The source also says RoboCurve reports that GPT-6.2 Astra tops its RoboDojo-RC Tier 1 real-robot benchmark, at about 70 times the cost of the robot model Dream-1. A separate claim says Artificial Analysis ranked it first on a benchmark, but the source is cut off before naming it. The source is a digest of these claims, so they have not been independently checked here.

AI-assisted summary based on the listed source.

OpenAI’s Thibault Sottiaux called the new models clearly capable but polarizing, with rough edges to sand down; Matt Shumer said the negative reviews reflect the model, not an embargo conspiracy, and Theo says Claude’s lead in T3 Code took no hit. RoboCurve says GPT-6.2 Astra tops its RoboDojo-RC Tier 1 real-robot...

Robotics teams and businesses weighing AI models for physical systems may care about the RoboCurve claim, since a top benchmark result comes with a cost about 70 times that of Dream-1. Buyers and builders will need to weigh that performance against the price gap. Next, watch for the full benchmark details from RoboCurve and Artificial Analysis, and for the caveat attached to the ARC-AGI-3 score.

Hardware and robotics watchers may want to track whether this becomes a product, benchmark, or deployment signal.

Signal Strength 91% Technical label WATCH Public Interest 37 Category ROBOTS & HARDWARE Reader Depth TECHNICAL

Signal Strength reflects source quality, relevance, freshness and evidence. Public Interest helps organize discovery; it is not proof of truth.

Public Interest components
Recognizable Entity Score 0 Practical Impact Score 18 Novelty Interest Score 94 Consequence Score 46 Curiosity Score 32 Shareability Score 48

VQV surfaced this signal because it is recent, relevant to Robotics, connected to Scobleizer News Pulse (X).