Command Palette

Search for a command to run...

Gemini 2.5 Flash-Lite (Reasoning)

Google

ReasoningVision ModelProprietary

Context

Knowledge Cutoff
Jan 01, 2025
Window
1M

PricingPer 1M tokens

Input
$0.1
Output
$0.4
Blended 3:1
$0.175

Capabilities

Speed
585 t/s
Input
Output
Reasoning tokens

Latency

TTFT
6.07 ms
500 token response
6.90 s

Benchmarks

Reasoning
●●●○○
Math
●●●●
Coding
●●○○○
MMLU Pro
75.9%
GPQA
63.0%
HLE
6.4%
SciCode
19.0%
AIME
70.0%
MATH 500
97.0%
LiveCodeBench
59.0%
HumanEval
97.0%

Gemini 2.5 Flash-Lite is Google’s stripped-down, multimodal transformer that handles text, images, audio, and short video while swallowing up to a million tokens. It prioritizes speed and price—responses start in ~6 ms and cost about $0.10 in, $0.40 out per million tokens—beating Gemini 1.5 Flash on most benchmarks without chasing GPT-4-level depth.

Use it when you need cheap, real-time output: customer-support chatbots, quick code fixes, bulk transcription, or other high-throughput tasks. Skip it for Olympiad-style math or heavyweight reasoning, but if latency and budget trump perfection, this is the pragmatic choice.