Alphabet stock falls as Google delays flagship Gemini 3.5 Pro model launch

Alphabet Inc. Shares fell on Thursday, 16 July 2026, following reports that Google has missed release targets for its flagship AI model, Gemini 3.5 Pro. Originally scheduled for a June 2026 launch as announced at the company’s I/O developer conference in May, the model remains unreleased as developers struggle to address performance shortcomings, particularly in the domain of coding.

Internal Performance and Development Hurdles

The postponement follows internal testing where the model failed to meet defined benchmarks. Google attempted to rectify these deficits by updating the model's training data late last month, yet reports indicate these efforts yielded disappointing results. This performance gap has fueled concern among staff, including engineers, researchers, and managers, who worry that Google is losing its competitive standing. The frustration has prompted some senior researchers to depart for competitors such as Anthropic.

Related imagery

Development has been further complicated by a fragmented internal structure. Multiple divisions—including Google Cloud, DeepMind, and the Android team—have been running independent, overlapping projects to create developer-facing coding tools. This lack of centralized direction has created competition for limited internal computing resources, leaving even internal users facing capacity constraints when attempting to utilize AI tools for their own tasks. Efforts to consolidate these efforts are underway; Chief AI Architect Koray Kavukcuoglu is working to unite internal coding tools, and a dedicated team under DeepMind, led by research engineer Sebastian Borgeaud, has been tasked with addressing these coding-specific challenges. Much of the company's internal developer tooling has been consolidated under the Antigravity platform, which manages safety, memory, and data protocols.

Philosophical and Competitive Pressures

Internal progress has also been affected by differing views on software quality. Some engineers maintain a purist philosophy, arguing that critical code must be written by humans to satisfy the company’s long-standing quality standards. This stance has periodically slowed the adoption of AI-generated code, although Google recently reported that 75 percent of its new code is now generated by AI and approved by engineers. Meanwhile, co-founder Sergey Brin has reportedly pressured teams to accelerate the development of coding capabilities to better compete with rivals.

The industry competitive landscape has shifted rapidly, with OpenAI and Meta Platforms Inc. Recently releasing models that outperform Google’s current offerings in code generation. Anthropic has also remained a primary competitor.

Market Reception and Interim Strategies

While the Pro model remains in development, Google has released the Gemini 3.5 Flash model for customer use. Reception has been mixed. Rodrigo Davies, a product manager at Figma, noted that the model achieved a favorable balance of speed and quality for his team’s needs. Conversely, Freddy Vega, CEO of the education platform Platzi, stated his firm shifted to Anthropic’s Claude 3.5 Sonnet, describing the Google Flash model as slower and more expensive than its predecessor while failing to match the performance of competitor products.

Google maintains that it is shipping models quickly across a broad portfolio. A company spokesperson stated that the firm is currently testing 3.5 Pro, an upgraded Flash model, and other systems with partners, while remaining in regular contact with the US government regarding safety standards and model testing.