Gemini 1.5 Flash

Google DeepMind

Gemini 1.5 Flash was released May 24, 2024. A lightweight, fast model distilled from 1.5 Pro. Designed for high-volume, cost-sensitive tasks. Featured a 1M token context window at an extremely low price point (\$0.075/\$0.30 per million tokens).

Source →

Developer

Google DeepMind

Released

2024-05-24

Context Window

1M tokens

Pricing

Type	Price per MTok
Input	$0.07
Output	$0.30

Benchmarks6

Knowledge

2 benchmarks

MMLU

78.9%#33/37

12th percentile

MMLU-Pro

67.3%#14/15

10th percentile

Reasoning

1 benchmark

BBH

85.5%#5/8

44th percentile

Math

2 benchmarks

MATH

77.9%#17/31

47th percentile

MGSM

82.6%#7/7

7th percentile

Coding

1 benchmark

HumanEval

74.3%#21/25

18th percentile

Percentile among tested models:Top 25%50-75%25-50%Bottom 25%

Gemini Family5

Model	Tier	Released	Input $/MTok
Gemini 2.5 Flash	flash	2025-05-20	$0.15
Gemini 2.5 Pro	pro	2025-03-25	$1.25
Gemini 2.0 Flash	flash	2024-12-11	$0.10
Gemini 1.5 Pro	pro	2024-02-15	$1.25
Gemini 1.0 Ultra	ultra	2024-02-08	—

Details

Model FamilyGemini

Tierflash

Generation1.5

Release Date2024-05-24

Context Window1M tokens

Open WeightNo

Modalitytext, image, audio, video

Capabilities4

tool-usevisionaudio-inputlong-context

Sources2

Google I/O 2024 Gemini Announcements

Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context (technical report)