Q: What did Google actually announce?

A: On 30 September, Google DeepMind announced Gemini 4 Argon, its new flagship model. The company reported 77.9% on the DeepSWE v1.1 coding test, first place on AutomationBench and a one-million-token output limit. Introductory pricing is $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 later. Access so far is limited to a group of cyber-defence partners. Paid API customers and AI Ultra subscribers are promised access “as soon as possible”, with no date.

Q: Why does this model matter so much for Alphabet?

A: The last four Alphabet reports named a competitive Gemini 4 as the clearest test of the AI thesis. It came months late, after a cancelled intermediate release and an August leadership change at DeepMind. Those doubts help explain why the shares fell about 16% from their 18 May peak.

Q: So is the test passed?

A: Only half of it. The model exists and its benchmark claims are strong. Google itself calls it comparable to the latest OpenAI and Anthropic models on key coding and cyber tests, which is a careful claim, not a clear win. The price matches OpenAI’s flagship, so Alphabet is competing on cost as much as capability. The shares rose before the open on 1 October, then closed 1.7% lower at $338.24.

Q: What does it do to earnings and the price levels?

A: Very little to earnings. By the report’s estimate, a strong rollout adds about $0.10 to forward EPS and a stalled one removes about $0.12, both under 1%. The effect shows up in the multiple instead, where each turn of forward P/E is worth about $13 a share. A broad, well-received release would lift the fair value toward $438. A release that stays gated, or a price war, would pull it toward $419. The central estimate stays at $428.

Q: Why does the rating stay at Buy?

A: Because the launch removes part of the downside story without yet proving the upside one. At $346.47 the shares sit inside the $340 to $365 buy range. The first target of $430 is 24% away and the $310 stop is 10.5% away. A $3.2 billion publisher ad-tech claim and a 5.31% ten-year yield add pressure without changing the call.

Q: What should readers watch?

A: A general-availability date, independent benchmark results, and third-quarter results, expected in late October or early November, where search growth, Cloud growth and early Argon adoption arrive together. A release still gated by then would count against the stock.

The scenario table, the sizing inputs and the full trade plan are in the complete Alphabet (GOOGL) report, available on our Reports page.