Google

Gemini 3.5 Flash-Lite

Made by Google DeepMind. Released 2026-07-21. Accessibility not recorded. Cheapest standard rate we hold: $2.50 per million output tokens, no context split.

What it costs

TierKindInOutCached inFrom
batchtext$0.15$1.25seller
flextext$0.15$1.25seller
prioritytext$0.54$4.50seller
standardtext$0.30$2.50seller

per million tokens, US dollars. * marks a rate from a single registry or two that disagree — the best available number rather than a confirmed one. Every other row is confirmed against a seller or a licensed index.

What it scores, by house

BenchmarkHouseScoreEvaluatedScaffold
chess puzzlesEpoch AI0.2±0.042026-08-06high
chess puzzlesEpoch AI0.2±0.042026-08-06minimal
chess puzzlesEpoch AI0.2±0.042026-08-06low
frontiermath tier 4 v2Epoch AI0.0±0.002026-08-02high
frontiermath tiers 1 3 v2Epoch AI0.3±0.032026-08-02high
gpqa diamondEpoch AI0.7±0.032026-08-06minimal
gpqa diamondEpoch AI0.8±0.032026-08-06low
gpqa diamondEpoch AI0.8±0.032026-08-06high
mystery game puzzlesEpoch AI0.1±0.032026-08-05minimal
mystery game puzzlesEpoch AI0.2±0.042026-08-05low
otis mock aime 2024 2025near its ceilingEpoch AI0.5±0.082026-08-06minimal
otis mock aime 2024 2025near its ceilingEpoch AI0.7±0.072026-08-06high
otis mock aime 2024 2025near its ceilingEpoch AI0.6±0.072026-08-06low
ale benchcollected by Epoch AI765high
arc aginear its ceilingcollected by Epoch AI0.2low
arc aginear its ceilingcollected by Epoch AI0.5high
arc aginear its ceilingcollected by Epoch AI0.3medium
arc aginear its ceilingcollected by Epoch AI0.1minimal
arc agi 2collected by Epoch AI0.0minimal
arc agi 2collected by Epoch AI0.0low
arc agi 2collected by Epoch AI0.1medium
arc agi 2collected by Epoch AI0.1high
critptcollected by Epoch AI0.0
dtbenchnear its ceilingcollected by Epoch AI0.8high
gdp pdfcollected by Epoch AI0.1high
lmcacollected by Epoch AI37.6high
proofbenchcollected by Epoch AI0.1
scicodecollected by Epoch AI0.4
webdev arenacollected by Epoch AI1449
weirdmlcollected by Epoch AI0.4high

These scores are not comparable between houses and are not added up. A score is dated by when it was evaluated, not when the model was released, and a score without its scaffold is not a measurement — which is why both are printed.

Every rate we hold · Why there is no single best model · Everything from Google