GoogleReasoning modelAuto-detected

Gemini 3.5 Flash Lite

Name: Gemini 3.5 Flash Lite benchmark results
Creator: BenchLeader

Best configuration ranks #221 of 610 on the BenchLeader Index at 53.5 ±8.1. Last measured 8 Sept 2026. Released 21 Jul 2026.

Blended price: $0.850/M; $0.300 in · $2.50 out
Output speed: 369 tok/s
First answer: 8.01 s; first token 1.07 s
Context: 1.0M

How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.

Overall index54
Reasoning65
Coding53
Agents & tools18
Maths38
Knowledge67
Human preference65
Multimodal63
Long context65
Composite58

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.

Effort	Index	Rank	Speed	First answer	Chat reply cost	Categories
minimal	–	–	369 tok/s	8.01 s	$0.0009	Maths 46 · Reasoning 41
low	–	–	369 tok/s	8.01 s	$0.0009	Maths 50 · Reasoning 42
medium	–	–	369 tok/s	8.01 s	$0.0009	Reasoning 37
high	44.0	#426	369 tok/s	8.01 s	$0.0009	Agents & tools 39 · Coding 52 · Composite 18 · Knowledge 57 · Maths 42 · Reasoning 49
defaultbest	53.5	#221	369 tok/s	8.01 s	$0.0009	Agents & tools 18 · Coding 53 · Composite 58 · Human preference 65 · Knowledge 67 · Long context 65 · Maths 38 · Multimodal 63 · Reasoning 65

Benchmark results

One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Reasoning

Benchmark	minimal	low	medium	high	default	Source
GPQA Diamond	74.2%#141	75.8%#131	–	83.3%#91	–	Epoch AI Benchmarking Hub
LMArena Hard Prompts	–	–	–	–	1476#60	LMArena
LiveBench Reasoningnot in index	–	–	–	60.2%#52	–	LiveBench
GPQA Diamond (AA)not in index	–	–	–	–	83.8%#151	Artificial Analysis
Humanity's Last Exam (AA)not in index	–	–	–	–	18.8%#169	Artificial Analysis
GPQA Diamond (Vals)not in index	–	–	–	83.8%#54	–	Vals AI
ARC-AGI-1	7.5%#170	17.0%#159	32.3%#145	53.5%#120	–	ARC Prize
ARC-AGI-2	0.8%#164	1.5%#151	5.3%#118	10.3%#105	–	ARC Prize

Coding

Benchmark	minimal	low	medium	high	default	Source
SciCode	–	–	–	–	40.9%#102	SciCode
WeirdML	–	–	–	39.0%#108	–	WeirdML
LMArena Coding	–	–	–	–	1503#59	LMArena
LMArena WebDev	–	–	–	–	1449#53	LMArena
LiveBench Codingnot in index	–	–	–	76.1%#33	–	LiveBench
SciCode (AA)not in index	–	–	–	–	41.3%#110	Artificial Analysis
LiveCodeBench	–	–	–	79.0%#72	–	Vals AI
IOI	–	–	–	26.2%#17	–	Vals AI
SWE-bench (Vals)not in index	–	–	–	75.0%#42	–	Vals AI

Agents & tools

Benchmark	minimal	low	medium	high	default	Source
LMArena Agent	–	–	–	–	-13.4#40	LMArena
LiveBench Agentic Codingnot in index	–	–	–	45.3%#41	–	LiveBench
Terminal-Bench 2.1 (Vals)	–	–	–	50.2%#48	–	Vals AI

Maths

Benchmark	minimal	low	medium	high	default	Source
FrontierMath Tiers 1–3	–	–	–	26.0%#64	–	Epoch AI Benchmarking Hub
FrontierMath Tier 4	–	–	–	0.0%#57	–	Epoch AI Benchmarking Hub
OTIS Mock AIME	51.1%#165	60.0%#146	–	71.1%#120	–	Epoch AI Benchmarking Hub
ProofBench	–	–	–	–	13.0%#47	Vals AI
LiveBench Mathematicsnot in index	–	–	–	73.7%#52	–	LiveBench

Knowledge

Benchmark	minimal	low	medium	high	default	Source
LiveBench Data Analysisnot in index	–	–	–	53.3%#52	–	LiveBench
AA-Omniscience	–	–	–	–	5.2#72	Artificial Analysis
MMLU-Pro	–	–	–	85.8%#51	–	Vals AI
LegalBench	–	–	–	84.1%#36	–	Vals AI
CorpFin	–	–	–	60.7%#65	–	Vals AI
TaxEval	–	–	–	72.6%#60	–	Vals AI

Instruction following

Benchmark	minimal	low	medium	high	default	Source	Trend
LiveBench Languagenot in index	–	–	–	71.8%#47	–	LiveBench

Human preference

Benchmark	minimal	low	medium	high	default	Source	Trend
LMArena Text	–	–	–	–	1457#56	LMArena

Multimodal

Benchmark	minimal	low	medium	high	default	Source	Trend
LMArena Vision	–	–	–	–	1269#34	LMArena
MMMU-Pro	–	–	–	–	79.0%#45	Artificial Analysis

Long context

Benchmark	minimal	low	medium	high	default	Source	Trend
AA-LCR	–	–	–	–	76.0%#102	Artificial Analysis

Composite

Benchmark	minimal	low	medium	high	default	Source
Epoch Capabilities Indexnot in index	–	–	–	–	145.1#71	Epoch AI Benchmarking Hub
LiveBench	–	–	–	63.9%#51	–	LiveBench
AA Intelligence Index	–	–	–	–	22.7#142	Artificial Analysis
Vals Indexnot in index	–	–	–	36.7#38	–	Vals AI

Where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

Provider	Speed	First token	Input $/M	Output $/M	Context	Quantisation
Google Vertex Flex	169 tok/s	6.74 s	$0.150	$1.25	1.0M	–
Google AI Studio	72 tok/s	0.77 s	$0.300	$2.50	1.0M	–
Google Vertex	68 tok/s	0.61 s	$0.300	$2.50	1.0M	–
Google AI Studio Priority	50 tok/s	1.12 s	$0.540	$4.50	1.0M	–
Google Vertex Priority	33 tok/s	1.07 s	$0.540	$4.50	1.0M	–
Google Vertex (EU)	23 tok/s	1.49 s	$0.330	$2.75	1.0M	–
Google AI Studio Flex	17 tok/s	0.70 s	$0.150	$1.25	1.0M	–

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.030 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0009	$0.0008	8.8 s
Summarise a 30-page report	12,000 / 600	$0.0051	$0.0027	9.6 s
Code edit	6,000 / 1,500	$0.0056	$0.0043	12.1 s
Agentic coding session	60,000 / 4,000	$0.028	$0.016	18.8 s
Structured extraction	2,000 / 200	$0.0011	$0.0007	8.6 s