VQV Signal

RESEARCH WATCH TECHNICAL

GPT-6.2 Astra tops 3D benchmarks, but Epoch finds uneven gains

According to a post on X by @ArtificialAnlys, Artificial Analysis tested GPT-6.2 Astra, currently #1 on its AA-3D leaderboard, on whether a language model driving Blender and video-as-code can compete with video generation models. Andon Labs says Astra ranks #1 on Blueprint-Bench 2 for understanding 3D space. Separate Epoch tests found uneven gains: +25 points on AI data-center research and +15 on graphic design, but +1 on data insights and -3 on research. The available snippet is cut off, so the full results are not shown.

Source: Scobleizer News Pulse (X) · x.com Found via Robert Scoble’s Scobleizer News Pulse · a post on X, not independently verified Published 2026-11-20T02:40:00+00:00 Detected 2026-11-20T19:02:19+00:00
View original source

According to a post on X by @ArtificialAnlys, Artificial Analysis tested GPT-6.2 Astra, currently #1 on its AA-3D leaderboard, on whether a language model driving Blender and video-as-code can compete with video generation models. Andon Labs says Astra ranks #1 on Blueprint-Bench 2 for understanding 3D space. Separate Epoch tests found uneven gains: +25 points on AI data-center research and +15 on graphic design, but +1 on data insights and -3 on research. The available snippet is cut off, so the full results are not shown.

AI-assisted summary based on the listed source.

Artificial Analysis tested GPT-6.2 Astra, the current #1 on its AA-3D leaderboard, on whether a language model driving Blender and video-as-code can compete with video generation models. Andon Labs says Astra ranks #1 on Blueprint-Bench 2 for understanding 3D space. Epoch's work tests found uneven gains: +25...

Teams choosing models for 3D reasoning or code-driven video work should not rely on a top leaderboard rank alone, since the reported Epoch results range from +25 to -3 points across tasks. Readers should check the full benchmark breakdowns and methods before drawing conclusions. The next step to watch is the complete Epoch results, which the snippet does not include.

Signal Strength 91% Technical label WATCH Public Interest 33 Category RESEARCH 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 8 Novelty Interest Score 94 Consequence Score 34 Curiosity Score 32 Shareability Score 46

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