Blab AI
Alphabet Delays Gemini 3.5 Pro Launch, Citing Coding Performance Gaps
Alphabet Inc. (NASDAQ: GOOGL, GOOG) has pushed back the release of its next‑generation Gemini 3.5 Pro model, originally slated for June. The decision follows an internal review that found the model’s coding capabilities fell short of the company’s internal benchmarks after a recent training‑data update.
The Gemini line, Google’s flagship family of large language models, debuted in December 2023 as a direct competitor to OpenAI’s GPT‑4 and Anthropic’s Claude. Gemini 3 launched in late 2025, and the 3.5 generation was expected to continue a rapid cadence: Gemini 3.5 Flash was unveiled at Google’s I/O developer conference in mid‑May, and Gemini 3.5 Pro was set to follow in June.
Bloomberg reported that Google refreshed the dataset used to train Gemini in an effort to tighten its coding performance. The results were underwhelming, prompting the company to postpone the Pro release. No new launch window has been announced.
In a statement, Google said it is "currently testing 3.5 Pro, an upgraded Flash model, and other models with partners," adding that it is "shipping quickly across a wide range of models while keeping them highly cost‑effective for customers." The company emphasized that coding remains a key battleground for enterprise customers.
The delay has sparked questions about Alphabet’s competitive standing. While internal metrics suggest the postponement may not immediately alter the company’s investment case, analysts warn that a flagship model that does not meet internal standards could erode confidence. Alphabet’s 2026 capital‑expenditure plan includes up to $190 billion for AI infrastructure, a figure that presumes the Gemini line remains market‑leading.
Shares of Alphabet slipped roughly 15% from a recent high of $408.61 to $346.77 on Friday—a modest pullback but a noticeable wobble for a stock that has long been a proxy for AI growth.
The broader AI arena is fiercely competitive. OpenAI continues to ship new models, and Anthropic has released Claude 3.5 Sonnet, which reportedly outperforms its larger sibling on certain benchmarks. In that context, a delay in Gemini 3.5 Pro could be viewed as a setback for Google.
Industry observers note that the timing of the delay is significant. The Gemini 3.5 generation was intended to sustain the momentum of the 3.5 release cycle, and a month‑long slip is a minor hiccup. However, repeated delays could signal deeper issues in Google’s model‑development pipeline.
Alphabet’s strategy, as outlined in its public filings, relies on a mix of on‑device models (Nano), cost‑effective real‑time models (Flash), and high‑compute models (Pro and Ultra). The company’s focus on coding performance is reflected in its investment in specialized training data and partner testing.
At present, Alphabet has not indicated a new release window for Gemini 3.5 Pro. It remains unclear whether the delay will extend beyond July 2026 or whether the model will ultimately meet the performance thresholds required for enterprise deployment.
The situation underscores the importance of reliable coding capabilities in the competitive AI market. As Alphabet continues to invest heavily in AI infrastructure, the outcome of the Gemini 3.5 Pro development will be closely watched by investors, enterprise customers, and competitors alike.
In summary, Alphabet has postponed the launch of Gemini 3.5 Pro to address coding performance gaps discovered during a recent training data update. The company has not set a new release date, and the delay may influence perceptions of Alphabet’s AI leadership, though its broader investment case remains largely intact for now.
The Gemini line, Google’s flagship family of large language models, debuted in December 2023 as a direct competitor to OpenAI’s GPT‑4 and Anthropic’s Claude. Gemini 3 launched in late 2025, and the 3.5 generation was expected to continue a rapid cadence: Gemini 3.5 Flash was unveiled at Google’s I/O developer conference in mid‑May, and Gemini 3.5 Pro was set to follow in June.
Bloomberg reported that Google refreshed the dataset used to train Gemini in an effort to tighten its coding performance. The results were underwhelming, prompting the company to postpone the Pro release. No new launch window has been announced.
In a statement, Google said it is "currently testing 3.5 Pro, an upgraded Flash model, and other models with partners," adding that it is "shipping quickly across a wide range of models while keeping them highly cost‑effective for customers." The company emphasized that coding remains a key battleground for enterprise customers.
The delay has sparked questions about Alphabet’s competitive standing. While internal metrics suggest the postponement may not immediately alter the company’s investment case, analysts warn that a flagship model that does not meet internal standards could erode confidence. Alphabet’s 2026 capital‑expenditure plan includes up to $190 billion for AI infrastructure, a figure that presumes the Gemini line remains market‑leading.
Shares of Alphabet slipped roughly 15% from a recent high of $408.61 to $346.77 on Friday—a modest pullback but a noticeable wobble for a stock that has long been a proxy for AI growth.
The broader AI arena is fiercely competitive. OpenAI continues to ship new models, and Anthropic has released Claude 3.5 Sonnet, which reportedly outperforms its larger sibling on certain benchmarks. In that context, a delay in Gemini 3.5 Pro could be viewed as a setback for Google.
Industry observers note that the timing of the delay is significant. The Gemini 3.5 generation was intended to sustain the momentum of the 3.5 release cycle, and a month‑long slip is a minor hiccup. However, repeated delays could signal deeper issues in Google’s model‑development pipeline.
Alphabet’s strategy, as outlined in its public filings, relies on a mix of on‑device models (Nano), cost‑effective real‑time models (Flash), and high‑compute models (Pro and Ultra). The company’s focus on coding performance is reflected in its investment in specialized training data and partner testing.
At present, Alphabet has not indicated a new release window for Gemini 3.5 Pro. It remains unclear whether the delay will extend beyond July 2026 or whether the model will ultimately meet the performance thresholds required for enterprise deployment.
The situation underscores the importance of reliable coding capabilities in the competitive AI market. As Alphabet continues to invest heavily in AI infrastructure, the outcome of the Gemini 3.5 Pro development will be closely watched by investors, enterprise customers, and competitors alike.
In summary, Alphabet has postponed the launch of Gemini 3.5 Pro to address coding performance gaps discovered during a recent training data update. The company has not set a new release date, and the delay may influence perceptions of Alphabet’s AI leadership, though its broader investment case remains largely intact for now.
