Gemini 4 Argon vs Claude Opus 5.5: three 3D tests, and how Google built the temple
From the reels
Reel 32 · 6 tools · October 16, 2026
Per token, Argon costs half of Opus until Nov 30, 2026, then the same. So a run's bill comes from how many tokens the model burns on that prompt, not from the price list alone.
How it works
- 1Blueprints first. The model called a tool to create architectural blueprints (CAD) before building anything.
- 2Four reference images. Nano Banana generated four images of the temple from different angles. They served as the references for everything after.
- 3A team of agents with a critic. Four Gemini 4 Argon specialist subagents worked in parallel, managed and reviewed by an Argon orchestrator acting as an adversarial critic.
- 4Compare, over and over. Renders from Blender were compared against the four reference images, again and again. George says he came back the next morning to the finished still renders.
Good to know
- The costs in the three tests are what each tester reported. None of them published how they measured it (tokens, settings, effort level).
- The four stages are what George says and shows in the video. Google has not published the full prompt or a skill file for it.
- "Not with 'build me a pagoda'" is a paraphrase of George's words: it was "a little more complex than just build me a pagoda".
- An independent guide by @copilot_shogo. Not affiliated with Google, Anthropic or the testers.
Free for people building with Claude Code
Two of us run a six‑figure business on this Claude Code setup.
Most setups forget who you are every session. This one knows your goal, pushes back when you drift, and checks its own work. Take the starter version. It takes 15 minutes to set up.
claude · your-notes/
✻
CLAUDE.md
- START-HERE.md
- CLAUDE.md
- prompts/interview.md
- agents/verifier.md
- diagnostic.md
Confirm your email and the download starts. One email a week after that, unsubscribe any time.
MORE LIKE THIS