Google Gemini 3.6 Flash: A New Frontier in AI Efficiency

On July 21, 2026, Google released Gemini 3.6 Flash, a reasoning model that delivers sustained frontier-level intelligence at higher speed and lower cost than any previous Flash model — designed explicitly for the agentic era of autonomous AI workflows.

The model supports a 1-million-token context window, equivalent to roughly 1,500 A4 pages of text, 30,000 lines of code, or an hour of video. It accepts text, image, speech, and video input, and outputs text with a maximum of 65,536 tokens per response. Independent benchmarks place its output speed at 304 tokens per second — the fastest among 186 comparable models — while its intelligence score of 50 on the Artificial Analysis Intelligence Index puts it well above the class average of 31.

Pricing is set at .50 per million input tokens and .50 per million output tokens, with a 90% cache-hit discount bringing cached input down to /usr/bin/bash.15 per million tokens. Google also released Gemini 3.5 Flash-Lite alongside it, a high-throughput variant optimized for agentic search and document processing at /usr/bin/bash.30 per million input tokens. Together, the new models are available immediately in the Gemini app, Google Antigravity, AI Studio, and Android Studio, with Gemini 3.5 Flash-Lite also rolling into Google Search.

Gemini 3.6 Flash excels at code generation, agentic execution, and spatial reasoning — three capabilities that are increasingly central to how developers build AI-native applications. Its reasoning variant, which this benchmark represents, chains through multi-step problems before producing final answers, making it particularly suited for complex software refactoring, data analysis, and automated workflow orchestration. The model's 1M context window also allows it to ingest and reason over entire codebases or lengthy documents in a single pass, reducing the need for chunking strategies that often degrade coherence.