Google DeepMind has announced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. The naming suggests a split between a standard live model and a variant that spends extra effort on reasoning before responding.
Because the source is a single announcement, no independent benchmarks or third-party assessments are available. The distinction between the two variants is presented by DeepMind itself, so the practical difference in behavior will need to be verified through testing.
The move continues the pattern of shipping multiple versions of a model for different use cases, though the source does not specify target applications, availability, or performance details.