Recent statements from Google DeepMind’s new chief Koray Kavukcuoglu highlight Gemini 4 entering early post-training with an intent to release an initial version “as soon as possible,” well before year-end, after prior flagship delays such as the unshipped Gemini 3.5 Pro. This cautious optimism amid competitive pressure from OpenAI and Anthropic updates underpins the market’s modest tilt toward a late-October or later launch, while the broad dispersion across October dates reflects uncertainty over post-training duration, internal testing, and typical Google rollout patterns via API and app. Traders weigh the absence of a confirmed October preview event or benchmark drop against the executive emphasis on accelerated iterations.