Google DeepMind’s July 21, 2026 release of Gemini 3.5 Flash-Lite alongside 3.6 Flash established the current Lite tier as its fastest, lowest-cost 3.5-class model, optimized for 350 output tokens per second and high-throughput agentic tasks like document processing and classification at $0.30/$2.50 per million tokens. Since then, Google has iterated the main Flash line with 3.7 and 3.8 variants for better coding and multimodal performance while keeping 3.5 Flash-Lite as the default for free users and volume workloads into October 2026. No 3.6+ Flash-Lite has shipped, as focus remains on token efficiency gains and agent scaling ahead of any Pro or Gemini 4 expansion. Competitive pressure from OpenAI and Anthropic lightweight models, plus internal emphasis on cost-sensitive production traffic, shapes expectations for the next Lite update.