Gemini 3.5 Flash
Sustained frontier-level intelligence for agentic execution, iterative coding cycles, multi-step tool use, finance analysis, and professional knowledge work — at Flash speed.
Empirical Evaluation Results
Architectural Profile & Capabilities
Sustained frontier-level intelligence for agentic execution, iterative coding cycles, multi-step tool use, finance analysis, and professional knowledge work — at Flash speed. Strongest agentic + coding Flash model Google has shipped; beats Gemini 3.1 Pro on every coding/agentic/multimodal benchmark Google evaluated, while running ~4× faster. Configurable thinking levels (minimal/low/medium/high; medium default). 90% cache discount on cached input ($0.15 / M cached). Released May 19, 2026 at Google I/O 2026. Knowledge cutoff January 2025. Computer Use not supported. Tradeoffs vs Gemini 3.1 Pro: regresses on pure abstract reasoning (Humanity's Last Exam: 40.2% vs 44.4%; ARC-AGI-2: 72.1% vs 77.1%) and 128K long-context retrieval (MRCR v2 8-needle: 77.3% vs 84.9%).
Sparse Mixture-of-Experts routing: activates only a subset of parameter experts per token, delivering frontier-level intelligence with exceptional efficiency.
Recommended Workloads & Primary Use Cases
Calls to Gemini 3.5 Flash are routed through strict Zero Data Retention inference channels. Prompts and outputs are never stored, indexed, or monitored by Google.
More Models from Google
Compare other engines available in this laboratory.
Gemini 3.1 Flash TTS is a neural speech synthesis model by Google, providing expressive voice rendering for real-time conversation.
Abstract reasoning puzzles, competitive algorithms, agentic tool orchestration, scientific coding, large document analysis (10K+ tokens), technical/academic writing
Legacy large-context analysis