With Gemini 3.8 Flash, Google reminds everyone it's still in the race

Google unveiled Gemini 3.8 Flash on Wednesday, aiming to reaffirm its standing as a leading model developer in the field. The company's dedication to the model race has been questioned since early August, when DeepMind CEO Demis Hassabis stepped down to become chairman of the AI research firm and Koray Kavukcuoglu took over leadership under a different title.

Google CEO Sundar Pichai's reassurances have been insufficient, particularly following the delayed release of Gemini 3.5 Pro in June and the lower benchmark metrics of Gemini 3.5 Flash compared to open-weight models from Chinese AI companies. However, Google has shown rapid iteration, with Gemini 3.8 Flash being its fourth Flash model in four months.

This new model has seen significant improvements over 3.7 Flash, often performing at a similar level to more expensive frontier models at a fraction of the cost. On the Artificial Intelligence Analysis Index, 3.8 Flash scored 59, matching the score of GPT-5.6 Sol and Grok 4.6. In terms of cost, it is the most affordable model at its intelligence level, being 40% more expensive per task than its predecessor.

The model's performance on complex tasks has also improved, showing greater diligence with extra reasoning steps and tool usage when needed. Currently, Gemini 3.8 Flash is available for developers through Google Antigravity, Gemini API, Google AI Studio, Android Studio, and Stitch. Enterprise users can utilize Gemini Enterprise, while consumers with Google AI Pro or Ultra subscriptions can access it via the Gemini app, AI Mode in Google Search, and Gemini in Google Sheets.

Additionally, a new program called Fairwind provides government, critical infrastructure, and software maintainers with access to the model for software vulnerability hunting and remediation.