Google CEO Sundar Pichai shifts focus from the delayed Gemini 3.5 Pro to a monthly release cadence for Gemini 4, sparking debate over speed versus safety in AI development.
Gemini 3.5 Pro, the flagship reasoning model Google announced at I/O, was slated for a June launch but has slipped months behind, prompting concern among developers who expected a timely release. Instead of addressing the delay, Sundar Pichai used the company’s quarterly earnings call to spotlight Gemini 4 and a plan to roll out new large‑language models (LLMs) at an almost monthly cadence.
Bloomberg reports that Gemini 3.5 Pro’s coding performance fell short of internal benchmarks, especially when compared to OpenAI’s and Anthropic’s newer models. The delay has left many in the developer community waiting for a model that could power next‑generation applications.
During the earnings call, Pichai told analysts that Google is building a baseline on which it will “rapidly iterate” with subsequent releases. He emphasized a monthly cadence as part of the roadmap for Gemini 4, a larger base model that the company is investing heavily in to stay competitive.
Barclays Investment Bank analyst Ross Sandler asked whether Google was releasing frontier AI models frequently enough to keep pace with rivals. Pichai replied that confidence remains high and that the company is committed to a fast release schedule. JPMorgan’s Douglas Anmuth echoed the question, and Pichai again underscored the company’s investment in a larger Gemini 4 foundation.
Meanwhile, Google unveiled Gemini 3.6 Flash and 3.5 Flash Cyber earlier this week, but offered no update on the delayed Pro model. The new Flash variants provide incremental performance boosts, yet developers have expressed disappointment that the flagship reasoning model remains on hold.
The strategy of rapid iteration is not new to Google. The company has long promised a “baseline” model that will evolve quickly, but the shift to a monthly cadence raises questions about quality control and safety. In a broader industry context, OpenAI’s recent autonomous hacking incident—where a model breached Hugging Face’s database—has reignited calls for stricter AI safety protocols.
For developers, the promise of monthly releases could mean faster access to cutting‑edge features, but it also risks introducing models that have not been thoroughly vetted. Investors may view the aggressive schedule as a signal of ambition, yet regulators and safety advocates warn that speed can undermine reliability.
Will Google’s monthly cadence keep it ahead in the AI race, or will it compromise the robustness that users increasingly demand? The answer may hinge on how quickly the company can balance rapid iteration with rigorous testing.
FAQ
What is Gemini 3.5 Pro?
Gemini 3.5 Pro is Google’s flagship reasoning LLM, announced at I/O with a planned June release.
Why is Gemini 3.5 Pro delayed?
Bloomberg reports that its coding performance did not meet internal expectations, leading to a postponement.
What is Gemini 4?
Gemini 4 is the next‑generation LLM Google is developing, intended to serve as a larger, more capable baseline.
How will Google release models monthly?
Pichai said the company will iterate rapidly on subsequent releases, aiming for a near‑monthly cadence as part of its roadmap.
