Nvidia ran Claude Opus 5 on ARC-AGI-3.
On its own: 30% — the top result of every model tested. Same model + a custom harness + a supervisor agent = 100%. The harness is the real lever. Databricks: the wrong harness can 2x your cost. Stop chasing the newest model. Own your
New benchmark scene: I gave Claude Code (Opus 5) one prompt
Recreate the Matrix lobby shootout in Three.js from an empty repo. It generated all textures, SFX & music itself (OpenAI, ElevenLabs, Suno) and self-verified with screenshots. Result: 62s demo, 31/31 tests, 1
Opus 5 + higgsfield one-shotted a 2026 xbox-quality platformer that does not exist
0:00 already mid-stride, no establishing shot 0:04 the ring gate under the arch 0:08 half a beat of hesitation at the canyon lip 0:12 the grapple swing every frame was generated. ( the exact
OPUS 5 MAKES a fully modular aircraft with 4000 PARTS and 150 FUNCTIONAL modules
Workflow - in comments ⬇️⬇️⬇️ It's around 80% finished, with entire mesh, parts, elements and everything visual you see has been procedurally coded and then sent to Blender by Opus 5. Avionics is
You have reached the end of the archive
All of Claude Opus 5