Google has internally registered model IDs for two new AI models — Gemini 3.6 Flash and Gemini 3.5 Flash Lite — while its flagship Gemini 3.5 Pro model is running behind schedule. The registrations, spotted in internal system logs, suggest the company is preparing interim releases to stay competitive in the fast-moving AI race.
Interim models in a fast-moving field
The new entries are seen as stopgap products rather than major leaps. Flash models typically offer faster, lighter performance at lower cost, while the Pro line is meant to deliver top-tier reasoning and capability. By pushing out a 3.6 Flash and a 3.5 Flash Lite, Google appears to be buying time until the full Pro version is ready.
Competitors have been releasing updates at a rapid clip. OpenAI, Anthropic, and others have shipped multiple model versions over the past year, putting pressure on Google to keep its Gemini lineup fresh. The Flash Lite variant could target mobile or edge devices where efficiency matters more than raw power.
What the model IDs suggest
Registration of model IDs does not guarantee a public launch. Companies often test multiple internal builds before deciding which ones to ship. Still, the naming convention — bumping the Flash to 3.6 while the Pro remains at 3.5 — indicates that the Flash line is evolving independently.
Industry watchers have noted that Google has been iterating faster on its smaller models. The 3.6 Flash could bring improvements in speed, context window, or instruction following without the cost of a full Pro upgrade. The 3.5 Flash Lite might be a stripped-down version for developers who need a cheap, fast API option.
Delays for the flagship Pro
The delay of Gemini 3.5 Pro is not entirely surprising. Large models often face hurdles in training stability, safety alignment, or meeting performance benchmarks. Google has not publicly explained the holdup, but the company has previously paused releases to address issues with factual accuracy or harmful outputs.
Without a flagship Pro model, Google risks falling behind in the premium AI tier. The company's cloud business and enterprise customers typically expect the most capable models for complex tasks. Interim Flash models can fill some gaps but may not satisfy users who need deep reasoning or multimodal capabilities.
Google has not commented on the registration or the delay. The model IDs were first flagged by developers monitoring internal infrastructure. Whether the 3.6 Flash and 3.5 Flash Lite see a public release remains unclear, but their existence shows the company is still iterating behind the scenes.
