AIbase
EN
Home
AI Models Finder
Gemini 2.5 Flash Lite
G

Gemini 2.5 Flash Lite

Google's lightest multimodal model optimized for high-throughput applications. Provides 250+ tokens/second ultra-fast inference and 0.25s time-to-first-token, supports multimodal input processing. Extremely cost-effective at only $0.075/million input tokens, particularly suitable for large-scale, high-frequency simple tasks and real-time interactive applications.
Intelligence(Medium)
Speed(Fast)
Input Supported Modalities
Yes
Is Reasoning Model
1,000,000
Context Window
65,536
Maximum Output Tokens
2025-01-01
Knowledge Cutoff
Go Compare

Pricing

￥0.72 /M tokens
Input
￥2.88 /M tokens
Output
￥1.26 /M tokens
Blended Price

Quick Simple Comparison

Input

Output

Gemini 2.0 Flash Thinking Experimental (Dec '24)
Gemini 1.5 Pro (Sep '24)
￥2.5
Gemini 2.0 Pro Experimental (Feb '25)

Basic Parameters

Gemini 2.5 Flash-LiteTechnical Parameters
Parameter Count
Not Announced
Context Length
1.0M tokens
Training Data Cutoff
2025-01-01
Open Source Category
Proprietary
Multimodal Support
Text, Image
Throughput
5
Release Date
2025-06-17
Response Speed
498.81,454 tokens/s

Benchmark Scores

Below is the performance of Gemini 2.5 Flash-Lite in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.
Intelligence Index
45.63
Large Language Model Intelligence Level
Coding Index
28.85
Indicator of AI model performance on coding tasks
Math Index
71.3
Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks
MMLU Pro
72.4
Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video
GPQA
47.4
Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions
HLE
3.7
The model's comprehensive average score on the Hugging Face Open LLM Leaderboard
LiveCodeBench
40
Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems
SciCode
17.7
The model's capability in code generation for scientific computing or specific scientific domains
HumanEval
92.7
Score achieved by the AI model on the specific HumanEval benchmark test set
Math 500 Score
92.6
Score on the first 500 larger, more well-known mathematical benchmark tests
AIME Score
50
An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase