●MODEL — Gemini 3.8 Flash reached general availability on September 2, the third Flash release in six weeks. Pricing holds at 3.7 Flash levels: $0.75 input and $3.75 output per MTok●PRICE — That introductory rate runs through December 31. From January 1, 2027 it becomes $1.50 and $7.50 per MTok, which is worth folding into next year's estimates now●BENCH — Google reports 54.9% on HLE-Verified and, on DeepSWE v1.1, results ahead of most larger frontier models, along with gains on the Vals Finance Agent V2 and Harvey Legal Agent benchmarks●EFFORT — 3.8 Flash works harder on hard problems, taking extra reasoning steps and calling tools iteratively, so token counts can rise. Where efficiency comes first, 3.7 Flash remains fully supported●CYBER — Gemini 3.8 Flash Cyber launched alongside it, tuned for vulnerability discovery and automated patching, and offered only to trusted defenders through the Fairwind Program●AUDIO — Lyria 3.5 entered public preview on September 3. It takes text and images as input and generates full-length tracks in 44.1 kHz stereo●MODEL — Gemini 3.8 Flash reached general availability on September 2, the third Flash release in six weeks. Pricing holds at 3.7 Flash levels: $0.75 input and $3.75 output per MTok●PRICE — That introductory rate runs through December 31. From January 1, 2027 it becomes $1.50 and $7.50 per MTok, which is worth folding into next year's estimates now●BENCH — Google reports 54.9% on HLE-Verified and, on DeepSWE v1.1, results ahead of most larger frontier models, along with gains on the Vals Finance Agent V2 and Harvey Legal Agent benchmarks●EFFORT — 3.8 Flash works harder on hard problems, taking extra reasoning steps and calling tools iteratively, so token counts can rise. Where efficiency comes first, 3.7 Flash remains fully supported●CYBER — Gemini 3.8 Flash Cyber launched alongside it, tuned for vulnerability discovery and automated patching, and offered only to trusted defenders through the Fairwind Program●AUDIO — Lyria 3.5 entered public preview on September 3. It takes text and images as input and generates full-length tracks in 44.1 kHz stereo