A · closed · Gemini

Gemini 3.1 Flash-Lite

Google

B · closed · Gemini

Gemini 3.6 Flash

Google

Replace A

Replace B

0:11

overall capability wins (2 ties across 13 scored). Gemini 3.6 Flash leads overall.

Reasoning0:6
Coding0:3
Arena0:1
Tool use0:1
Composite indices0:2
Pricing3:0
Speed2:0

Verdict

Gemini 3.6 Flash leads coding (widest gap: +283.0 on WebDev Elo); Gemini 3.6 Flash ranks higher on LMArena (+48 Elo); Gemini 3.1 Flash-Lite is ~5.3x cheaper on a blended token basis than Gemini 3.6 Flash.

Gemini 3.6 Flash leads coding (widest gap: +283.0 on WebDev Elo)
Gemini 3.6 Flash ranks higher on LMArena (+48 Elo)
Gemini 3.1 Flash-Lite is ~5.3x cheaper on a blended token basis than Gemini 3.6 Flash

Cost for 1M in + 250K out

Illustrative chat workload at primary-provider list prices.

Gemini 3.1 Flash-Lite: $0.63
Gemini 3.6 Flash: $3.38

Gemini 3.1 Flash-Lite is about 5.4x cheaper on this workload.

Price per 1M tokens

Full comparison

Organization

Google

Family

Gemini

License

Proprietary

Open weights

Release

May 7, 2026

Jul 21, 2026

Knowledge cutoff

API / provider

Google

Modalities

text, image, audio, video → text

Specs

Context window

tie

Max output

66K

tie

Parameters

—

Pricing

Input $/1M

$0.25

$1.50

A +$1.25

Output $/1M

$1.50

$7.50

A +$6

Blended $/1M (3∶1)

$0.56

A +$2.44

Speed

tok/s

240

180

A +60

TTFT (s)

0.15

0.22

A +0.07

Reasoning

MMLU-Pro

—

GPQA Diamond

82.2

92.8

B +10.6 pts

Humanity's Last Exam

17.2

40.8

B +23.6 pts

AIME 2025

—

MATH-500

—

Humanity's Last Exam (with tools)

—

AA-Omniscience Accuracy

36.3

50.0

B +13.7 pts

AA-LCR v1.1

74.3

80.0

B +5.7 pts

CritPt

1.1

10.6

B +9.5 pts

MMMU-Pro

75.5

83.2

B +7.7 pts

IFBench

77.2

—

Chartography

—

Chartography (With Tools)

—

Coding

SWE-bench Verified

—

SWE-bench Pro

—

SWE-bench Multilingual

—

LiveCodeBench

—

Terminal-Bench 2.1

31.1

77.5

B +46.4 pts

Aider Polyglot

—

Terminal-Bench 3

—

BigCodeBench

—

SciCode

43.4

53.4

B +10.0 pts

CursorBench

—

53.5

—

SWE-Rebench

—

NL2Repo-Bench

—

DeepSWE

—

47.0

—

WebDev Arena

1,254

1,537

B +283 Elo

Terminal-Bench 4.0

—

CursorBench 4.0

—

FrontierCode v1.1 (Main)

—

Arena

LMArena Elo

1,432

1,480

B +48 Elo

Metric	Gemini 3.1 Flash-Lite	Gemini 3.6 Flash	Delta
Identity
Organization	Google	Google	—
Family	Gemini	Gemini	—
License	Proprietary	Proprietary	—
Open weights	No	No	—
Release	May 7, 2026	Jul 21, 2026	—
Knowledge cutoff	-	-	—
API / provider	Google	Google	—
Modalities	text, image, audio, video → text	text, image, audio, video → text	—
Specs
Context window	1M	1M	tie
Max output	66K	66K	tie
Parameters	—	—	—
Pricing
Input $/1M	$0.25	$1.50	A +$1.25
Output $/1M	$1.50	$7.50	A +$6
Blended $/1M (3∶1)	$0.56	$3	A +$2.44
Speed
tok/s	240	180	A +60
TTFT (s)	0.15	0.22	A +0.07
Reasoning
MMLU-Pro	—	—	—
GPQA Diamond	82.2	92.8	B +10.6 pts
Humanity's Last Exam	17.2	40.8	B +23.6 pts
AIME 2025	—	—	—
MATH-500	—	—	—
Humanity's Last Exam (with tools)	—	—	—
AA-Omniscience Accuracy	36.3	50.0	B +13.7 pts
AA-LCR v1.1	74.3	80.0	B +5.7 pts
CritPt	1.1	10.6	B +9.5 pts
MMMU-Pro	75.5	83.2	B +7.7 pts
IFBench	77.2	—	—
Chartography	—	—	—
Chartography (With Tools)	—	—	—
Coding
SWE-bench Verified	—	—	—
SWE-bench Pro	—	—	—
SWE-bench Multilingual	—	—	—
LiveCodeBench	—	—	—
Terminal-Bench 2.1	31.1	77.5	B +46.4 pts
Aider Polyglot	—	—	—
Terminal-Bench 3	—	—	—
BigCodeBench	—	—	—
SciCode	43.4	53.4	B +10.0 pts
CursorBench	—	53.5	—
SWE-Rebench	—	—	—
NL2Repo-Bench	—	—	—
DeepSWE	—	47.0	—
WebDev Arena	1,254	1,537	B +283 Elo
Terminal-Bench 4.0	—	—	—
CursorBench 4.0	—	—	—
FrontierCode v1.1 (Main)	—	—	—
Arena
LMArena Elo	1,432	1,480	B +48 Elo