GoogleReasoning modelAuto-detected

Gemini 3.5 Flash

Name: Gemini 3.5 Flash benchmark results
Creator: BenchLeader

Best configuration ranks #39 of 610 on the BenchLeader Index at 63.9 ±2.7 (medium reasoning effort). Last measured 8 Sept 2026. Released 19 May 2026.

Blended price: $3.38/M; $1.50 in · $9.00 out
Output speed: 209 tok/s
First answer: 17 s; first token 1.66 s
Context: 1M

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

Overall index64
Reasoning67
Coding59
Agents & tools68
Knowledge74
Instruction following72
Human preference67
Multimodal68
Long context64
Composite71

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	57.3	#142	187 tok/s	1.02 s	$0.0033	Agents & tools 74 · Composite 59 · Instruction following 49 · Knowledge 64 · Long context 57 · Maths 59 · Multimodal 65 · Reasoning 49
low	–	–	209 tok/s	17 s	$0.0033	Maths 63 · Reasoning 64
mediumbest	63.9	#39	209 tok/s	17 s	$0.0033	Agents & tools 68 · Coding 59 · Composite 71 · Human preference 67 · Instruction following 72 · Knowledge 74 · Long context 64 · Multimodal 68 · Reasoning 67
high	60.4	#86	209 tok/s	17 s	$0.0033	Agents & tools 64 · Coding 61 · Composite 50 · Human preference 68 · Knowledge 64 · Maths 57 · Multimodal 67 · Reasoning 62
default	60.7	#78	209 tok/s	17 s	$0.0033	Agents & tools 57 · Composite 71 · Human preference 33 · Instruction following 74 · Knowledge 74 · Long context 63 · Maths 58 · Multimodal 69 · Reasoning 72

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	86.4%#66	88.9%#45	–	92.8%#19	–	Epoch AI Benchmarking Hub
SimpleBench	–	–	–	–	76.7%#6	SimpleBench
LMArena Hard Prompts	–	–	1493#36	1498#29	–	LMArena
LiveBench Reasoningnot in index	–	–	–	82.0%#36	–	LiveBench
GPQA Diamond (AA)not in index	82.8%#163	–	92.1%#39	–	92.2%#38	Artificial Analysis
Humanity's Last Exam (AA)not in index	24.1%#138	–	41.3%#47	–	42.7%#36	Artificial Analysis
GPQA Diamond (Vals)not in index	–	–	–	92.7%#14	–	Vals AI
EnigmaEval	–	–	–	25.4%#4	–	Scale AI SEAL
ARC-AGI-1	48.8%#123	–	–	92.5%#35	–	ARC Prize
ARC-AGI-2	8.9%#108	–	–	72.1%#34	–	ARC Prize

Coding

Benchmark	minimal	low	medium	high	default	Source
SWE-bench Verified (Epoch)	–	–	–	79.3%#3	–	Epoch AI Benchmarking Hub
SciCode	–	–	–	53.1%#45	–	SciCode
WeirdML	–	–	–	62.6%#41	–	WeirdML
LMArena Coding	–	–	1503#56	1508#51	–	LMArena
LMArena WebDev	–	–	1491#42	1500#40	–	LMArena
LiveBench Codingnot in index	–	–	–	78.2%#22	–	LiveBench
SciCode (AA)not in index	–	–	–	–	53.9%#45	Artificial Analysis
LiveCodeBench	–	–	–	87.6%#12	–	Vals AI
SWE-bench (Vals)not in index	–	–	–	78.8%#28	–	Vals AI

Agents & tools

Benchmark	minimal	low	medium	high	default	Source
LiveBench Agentic Codingnot in index	–	–	–	49.0%#33	–	LiveBench
Terminal-Bench Hard	46.2%#30	–	39.4%#56	–	40.9%#51	Artificial Analysis
τ²-Bench Telecom (AA)not in index	58.8%#177	–	95.6%#20	–	95.3%#24	Artificial Analysis
Terminal-Bench 2.1 (Vals)	–	–	–	74.2%#15	–	Vals AI
MCP Atlas	–	–	–	83.6%#4	–	Scale AI SEAL
HiL-Bench	–	–	–	–	27.7%#12	Scale AI SEAL

Maths

Benchmark	minimal	low	medium	high	default	Source
FrontierMath Tiers 1–3	–	–	–	62.8%#26	–	Epoch AI Benchmarking Hub
FrontierMath Tier 4	–	–	–	26.8%#31	–	Epoch AI Benchmarking Hub
OTIS Mock AIME	80.0%#99	88.9%#61	–	95.6%#29	–	Epoch AI Benchmarking Hub
ProofBench	–	–	–	31.0%#28	–	Vals AI
LiveBench Mathematicsnot in index	–	–	–	88.2%#30	–	LiveBench
AIME 2026	–	–	–	–	95.0%#18	MathArena
HMMT February 2026	–	–	–	–	95.5%#6	MathArena
MathArena Apex	–	–	–	–	32.3%#9	MathArena

Knowledge

Benchmark	minimal	low	medium	high	default	Source
SimpleQA Verified	–	–	–	66.2%#9	–	Epoch AI Benchmarking Hub
LiveBench Data Analysisnot in index	–	–	–	64.9%#46	–	LiveBench
AA-Omniscience	0.7#91	–	20.8#40	–	21.2#38	Artificial Analysis
MMLU-Pro	–	–	–	89.5%#10	–	Vals AI
LegalBench	–	–	–	83.6%#47	–	Vals AI
CorpFin	–	–	–	64.7%#35	–	Vals AI
TaxEval	–	–	–	74.4%#36	–	Vals AI

Instruction following

Benchmark	minimal	low	medium	high	default	Source	Trend
LiveBench Languagenot in index	–	–	–	84.6%#11	–	LiveBench
IFBench	47.3%#186	–	74.6%#35	–	76.3%#21	Artificial Analysis

Human preference

Benchmark	minimal	low	medium	high	default	Source	Trend
LMArena Text	–	–	1476#27	1479#23	–	LMArena
EQ-Bench 4	–	–	–	–	1087#23	EQ-Bench

Multimodal

Benchmark	minimal	low	medium	high	default	Source	Trend
LMArena Vision	–	–	1306#9	1311#8	–	LMArena
MMMU-Pro	80.1%#36	–	83.9%#15	–	84.3%#12	Artificial Analysis

Long context

Benchmark	minimal	low	medium	high	default	Source	Trend
AA-LCR	61.3%#210	–	74.3%#116	–	73.3%#129	Artificial Analysis

Composite

Benchmark	minimal	low	medium	high	default	Source
Epoch Capabilities Indexnot in index	–	–	–	–	154.7#26	Epoch AI Benchmarking Hub
LiveBench	–	–	–	74.6%#29	–	LiveBench
AA Intelligence Index	23.9#126	–	33.6#65	–	33.0#67	Artificial Analysis
Vals Indexnot in index	–	–	–	53.1#24	–	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 AI Studio Priority	192 tok/s	2.13 s	$2.70	$16.20	1.0M	–
Google AI Studio	187 tok/s	1.75 s	$1.50	$9.00	1.0M	–
Google AI Studio Flex	158 tok/s	1.41 s	$0.750	$4.50	1.0M	–
Google Vertex	127 tok/s	1.57 s	$1.50	$9.00	1.0M	–
Google Vertex Priority	36 tok/s	1.48 s	$2.70	$16.20	1.0M	–
Google Vertex Flex	4 tok/s	13 s	$0.750	$4.50	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 $0.150 per 1M. Reasoning tokens are not modelled.

Workload	Tokens in / out	Cost	With caching	Time
Chat reply	400 / 300	$0.0033	$0.0029	18.4 s
Summarise a 30-page report	12,000 / 600	$0.023	$0.011	19.8 s
Code edit	6,000 / 1,500	$0.022	$0.016	24.1 s
Agentic coding session	60,000 / 4,000	$0.126	$0.065	36.1 s
Structured extraction	2,000 / 200	$0.0048	$0.0028	17.9 s