Back to Explore
Google: Gemini 2.5 Flash Lite Preview 09-2025
Google·Released 2025-09-25
1.0M ctxMoEMultimodal
About
Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better performance...
Related Models
Google: Gemini 3.5 Flash2026-05-19Gemini 3.5 Flash (medium)2026-05-19Gemini 3.5 Flash (minimal)2026-05-19Google: Gemini 3.1 Flash Lite2026-05-07Google: Gemini 3.1 Flash Lite Preview2026-03-03Google: Gemini 3.1 Pro Preview Custom Tools2026-02-25Google: Gemini 3.1 Pro Preview2026-02-19Gemini 3 Deep Think2026-02-05
Pricing
Input
$0.10
per 1M tokens
Output
$0.40
per 1M tokens
Blended
$0.17
per 1M tokens
Cheaper than 72% of models. Median price is $0.54/1M tokens.
Cost Calculator
Tokens per day1M
100K100M
Daily
$0.17
Monthly
$5.25
vs. Similar Models
Gemma 4 12B (Non-reasoning)Q:+0.1
$0.15-14%
Qwen3 VL 30B A3B (Reasoning)Q:+0.2
$0.34+93%
QwQ 32BQ:+0.3
$0.74+326%
Qwen3 235B A22B (Reasoning)Q:+0.3
$2.63+1400%
Performance
353
tokens/sec
Faster than 98% of models
0.43
seconds
Faster than 92% of models
0.43
seconds
Faster than 95% of models
Market Median
94 tok/s
275% faster
Median TTFT
1.11s
61% faster
Throughput/Dollar
2020
tok/s per $/1M
Speed Comparison
Gemini 2.5 Flash-Lite Preview (Sep '25) (Reasoning)
347 tok/s-2%
Granite 3.3 8B (Non-reasoning)
362 tok/s+2%
gpt-oss-120b (low)
340 tok/s-4%
Context Window
1.0M
tokens
Larger than 90% of models
Max Output
66K
tokens
6% of context
Benchmarks
MMLU-Pro
79.6%
GPQA Diamond
65.1%
HLE
4.6%
LiveCodeBench
64.1%
SciCode
28.5%
TerminalBench Hard
7.6%
MATH-500Not evaluated
AIMENot evaluated
AIME 2025
46.7%
IFBench
41.8%
Long Context Recall
48.0%
Tau2
30.4%
Market AverageTop Score