Gemini 3.5 Flash

Launched at I/O 2026 (May 19) as the new default model in the Gemini app, AI Mode Search, and Antigravity 2.0. 1M-token context, configurable Thinking Mode, multimodal in (text/image/audio/video) and text out. Google reports it beats Gemini 3.1 Pro on agentic, coding, and multimodal benchmarks at ~4x the output speed of peer frontier models. Self-reported benchmarks include Terminal-Bench 2.1 76.2%, GDPval-AA 1656 Elo, MCP Atlas 83.6%.

AA Intelligence Index v4.0: 55 (#8 of 151). 190.2 tok/s output speed (#3 overall). Pricing $1.50 / $9.00 per 1M input/output. First-token latency is notably high (22.5s vs 2.6s median) — the speed advantage is in steady-state generation rather than time-to-first-token. Gemini 3.5 Pro was previewed for June.