modelbenchmark.io

Gemini 3.5 Flash

Google · gemini-3-5-flash

Compare

Fast Gemini model balancing multimodal reasoning, tool use, and cost

Specification

google listing

Context
1M
Max output
66K
Released
2026-05-19
Knowledge cutoff
2025-01
Retires
2027-05-19
Open weights
no
Input
text, image, video, audio, pdf
Output
text

Price

US dollars per million tokens · google list price

Input
$1.50
Output
$9.00
Cache read
$0.15
Cache write
$0.0833

Batch and priority prices

US dollars per million tokens · per listing

PriceListing$/M
Batch inputvertex_ai$0.75
Batch outputvertex_ai$4.50
Priority inputdatabricks$3.375
Priority inputvertex_ai$2.70
Priority outputdatabricks$20.25
Priority outputvertex_ai$16.20

Batch input is 50% below the interactive input price at the same listing.

Rate limits

as listed

Requests per minute
2,000
Tokens per minute
800,000

Quality

14 benchmarks · 3 sources

Composite
80th
percentile of 344 scored models
Rank
71 / 344
±0.06 sd
Evidence
14 × 3
3 or more benchmarks, from 2 or more sources
Effort range
7.5
points between effort settings
coding80th1/3 bench
reasoning67th4/4 bench
math76th6/7 bench
knowledge85th1/1 bench
general46th1/1 bench

By reasoning effort

every setting placed on the same scale as the leaderboard

SettingCompositeBenchmarks behind it
medium88th1
low87th4
high83rd13
minimal81st3

Every score

one row per source and configuration — nothing averaged away

BenchmarkScoreConfigurationSourceRun
FrontierMath-v159.5 ±13.3high · 2 runsEpoch AI2026-06-05
SimpleQA Verified66.2 ±1.5highEpoch AI2026-08-27
GPQA diamond92.8 ±1.6highEpoch AI2026-05-22
↳88.9 ±2.2lowEpoch AI2026-08-06
↳86.4 ±2.4minimalEpoch AI2026-07-15
Settings differ by 6.4 points on this benchmark.
SWE-Bench verified79.3 ±1.8highEpoch AI2026-06-01
↳71.8medium · mini-SWE-agentSWE-bench—
Sources and settings disagree by 7.5 points. Both figures stand.
OTIS Mock AIME 2024-202595.6 ±2.7highEpoch AI2026-05-25
↳88.9 ±4.7lowEpoch AI2026-08-06
↳80.0 ±6.0minimalEpoch AI2026-07-15
Settings differ by 15.6 points on this benchmark.
Chess Puzzles45.0 ±5.0highEpoch AI2026-10-13
↳42.0 ±5.0lowEpoch AI2026-10-13
↳27.0 ±4.5minimalEpoch AI2026-10-13
Settings differ by 18.0 points on this benchmark.
FrontierMath-Tiers-1-3-v262.8 ±2.9highEpoch AI2026-06-10
Mystery Game Puzzles32.0 ±4.7highEpoch AI2026-07-27
↳28.0 ±4.5lowEpoch AI2026-08-05
Settings differ by 4.0 points on this benchmark.
LiveBench75.4high · 23 runsLiveBench2026-06-25
FrontierMath-Tier-4-2025-07-017.3 ±0.0high · 2 runsEpoch AI2026-06-05
FrontierMath-Tier-4-v226.8 ±7.0highEpoch AI2026-06-10
EBR-bench4.8highEpoch AI2026-06-25
OEIS Open Lite29.0 ±4.6highEpoch AI2026-08-11
OEIS Open22.1 ±1.9highEpoch AI2026-08-11

The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.

Available from 43 hosts

HostIn $/MOut $/MCache rdCache wrContextOutputRetires
googlevendor$1.50$9.00$0.15—1M66K—
unorouter$0.1857$1.1142——1M66K—
aihubmix$0.75$3.75$0.075—1M64K—
kilo$0.75$4.50$0.075$0.04171M66K—
302ai$1.50$9.00——1M66K—
abacus$1.50$9.00$0.15—1M66K—
cloudflare-ai-gateway$1.50$9.00$0.15—1M66K—
commandcode$1.50$9.00$0.15—1M66K—
crossmodel$1.50$9.00$0.15$1.501M66K—
databricks
$1.50
$1.875
$9.00
$11.25
$0.15
$0.1875
$1.8751M66K—
deepinfra$1.50$9.00——1M1M—
edenai$1.50$9.00$0.15$0.08331M66K—
fastrouter$1.50$9.00$0.15—1M66K—
gemini$1.50$9.00$0.15—1M66K—
github-copilot$1.50$9.00$0.15—200K64K—
google-vertex$1.50$9.00$0.15—1M66K—
impossibl$1.50$9.00$0.15—1M66K—
llmgateway$1.50$9.00$0.15$0.08331M66K—
llmgateway-providers · google-vertex$1.50$9.00$0.15$0.08331M66K—
merge-gateway$1.50$9.00$0.15—1M66K—
nano-gpt$1.50$9.00$0.15$0.08331M66K—
nearai$1.50$9.00$0.15—1M66K—
neon$1.50$9.00$0.15—1M66K—
ofox$1.50$9.00$0.15$0.0831M66K—
opencode$1.50$9.00$0.15—1M66K—
openrouter$1.50$9.00$0.15$0.08331M66K—
opper$1.50$9.00$0.15—1M66K—
orcarouter$1.50$9.00$0.15$0.08331M66K—
perplexity$1.50$9.00$0.15————
prime$1.50$9.00——1M66K—
requesty$1.50$9.00$0.15$1.5831M66K—
sap-ai-core$1.50$9.00$0.15—1M66K—
tempr$1.50$9.00$0.15—1M66K—
vercel$1.50$9.00$0.15—1M64K—
vertex_ai$1.50$9.00$0.15—1M66K2027-05-19
vertex_ai-language-models$1.50$9.00$0.15—1M66K2027-05-19
zenmux$1.50$9.00$0.15—1M66K—
poe$1.5152$9.0909$0.1515—1M66K—
venice$1.55$9.45$0.155$0.0861M66K—
xpersona$1.55$12.20$0.155—1M128K—
cortecs$1.649$9.899$0.165$1.001M66K—
kyma · kyma$2.025$12.15$0.2025—1M8K—
kenari————1M66K—

Price history

append-only observations · a listing writes a row only when its price moves

DateHostField$/M
2026-09-02aihubmixCache read$1.50
2026-10-22aihubmixCache read$0.075
2026-09-02openrouterCache read$0.15
2026-09-06openrouterCache read$0.15
2026-10-10openrouterCache read$0.15
2026-09-02openrouterCache write$0.0833
2026-09-19openrouterCache write$0.0833
2026-10-10openrouterCache write$0.0833
2026-09-02aihubmixInput$1.50
2026-10-22aihubmixInput$0.75
2026-09-02openrouterInput$1.50
2026-09-06openrouterInput$1.50
2026-10-10openrouterInput$1.50
2026-09-02aihubmixOutput$9.00
2026-10-22aihubmixOutput$3.75
2026-09-02openrouterOutput$9.00
2026-09-06openrouterOutput$9.00
2026-10-10openrouterOutput$9.00
2026-09-06databricksCache read$0.1875
2026-10-17databricksCache read$0.15
2026-09-06databricksInput$1.875
2026-10-17databricksInput$1.50
2026-09-06databricksOutput$11.25
2026-10-17databricksOutput$9.00

More from Google

most-hosted first