Google said Gemini 3.6 Flash improves efficiency and coding performance at Gemini 3.5 Flash pricing, while Gemini 3.5 Flash-Lite targets high-concurrency workloads as Sundar Pichai defends the broader Gemini roadmap.
Google released Gemini 3.6 Flash and Gemini 3.5 Flash-Lite on July 22, adding a more efficient Flash model and a faster option for high-concurrency workloads. Google said Gemini 3.6 Flash improves output quality at the same price as Gemini 3.5 Flash, reduces average token consumption by about 17%, cuts usage by as much as 65% in DeepSWE scenarios, and lowers average agent-task time from 2.7 minutes to 1.3 minutes while reducing per-task cost from $0.59 to $0.50. CEO Sundar Pichai also said on Alphabet’s earnings call that Gemini 3.6 Flash improved by more than 10 points on a coding benchmark while using fewer tokens, as he highlighted the Flash line for cybersecurity, customer-service agents, data analytics and enterprise software. Gemini 3.5 Flash-Lite is positioned for high-concurrency tasks with output speeds of up to 350 tokens per second. The launch came as investors focused on delays to Gemini 3.5 Pro, which Pichai said remains in partner testing after missing its planned June debut, and on his comment that Gemini 4 releases could come at almost a monthly cadence.