BREAKING
Google Ships Gemini 3.6 Flash & 3.5 Flash-Lite
0
%
fewer output tokens
0
$
per M output tokens
0
$
per M input tokens
3.6 Flash Benchmark Scores
SWE-Bench Pro
58.7
OSWorld
83
CharXiv
89.4
Two Models, Two Jobs
3.6 Flash
1M context
●
Leaner output, lower cost
●
Multimodal, tool use, code
3.5 Flash-Lite
~350 tok/s
●
Fastest in the series
●
Parallel sub-agents, bulk docs
0
%
Harvey drafting speedup
0
%
JetBrains task gains
Betting on Token Efficiency
AI NEWS BLITZ
Google has launched two new Gemini Flash models built for running AI agents at scale.