What is being reviewed
- Theo describes Gemini 4 Argan as Google's first large Pro-like model release in eight months, following Gemini 3.1 Pro in February. He contrasts that interval with the one-to-two-month gap between the competing releases he names.
0:00 - The review compares pricing, front-end output, 3D games, specialized capabilities, and code work against Opus 5.5, 6.1 Soul, Astra, and Muse variants.
0:22 - Google contacts arranged early access. Theo says Google did not pay for this Gemini or Antigravity coverage, while disclosing that YouTube pays him as a creator.
0:511:09
Availability and vendor evidence
- The launch announcement lists API, Studio, and Gemini-platform access for developers and enterprises; Ultra access in Antigravity; and Ultra/Pro access in the Gemini app and AI Mode in Search. This is availability as described in the video.
2:38 - Google claims a number-one ranking on the "VL index" across economic domains. Theo challenges the comparison's value calculation because it assigns Muse Wonderful a token price he says was not available.
3:093:37 - Google reports over 35% more agentic code submissions and nearly one-third fewer rollbacks. Its examples include a refactor across hundreds of files completed in weeks rather than months and Python optimization in hours. These are vendor claims, not Theo's reproduced tests.
5:456:40 - A customer testimonial claims 100% completion on long-running workflows. Theo's objection is methodological: that percentage means little without knowing the tasks and their relevance to a user's work.
7:127:43