Google DeepMind · closed · 2026-03

Gemini 3.1 Flash-Lite

Item: Gemini 3.1 Flash-Lite
Rating: 83.4

#11 index 83.4

the take
Cheapest and fastest thing here, and it punches well above the price on knowledge. Do not hand it the hard reasoning; that is not the job it was built for.

benchmarks

—

86.9

88.9

—

LiveCodeBench (coding)

72.0

specs

Context: 1.048576M
Input: $0.25/M
Output: $1.5/M
Speed: 310.7 tok/s
Modality: text, image, audio, video

strengths

Dirt cheap ($0.25/$1.50), ~311 tok/s
Strong knowledge for the price
1M multimodal context

weaknesses

Weakest hard reasoning in the family
No published SWE-bench
Text-only output

Sources: [1][2]

Back to leaderboard Battle Gemini 3.1 Flash-Lite