Gemini 3 Flash Preview

ModelActive

by Google · family “gemini-flash-preview”

Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance. It delivers near Pro level reasoning and tool... (description from the OpenRouter listing)

in: textin: imagein: filein: audioin: videoout: textReasoningTool useStructured output

Epoch Capabilities Index

Not scored by Epoch AI

Listing ↗

Released

17 Dec 2025

Context window

1.05M tokens

Max output

66K tokens

Input $ / 1M tokens

$0.50

Output $ / 1M tokens

$3.00

Cached input $ / 1M

$0.050

Each benchmark shown separately with its own source

Benchmark results (23)

All benchmarks

Benchmark	Domain	Score	vs best recorded	Setting	Run	Source
LMCA	agents	43.1%	66%	high	—	External ↗
DTBench	reasoning	89.1%	91%	high	—	External ↗
Mystery Game Puzzles	games	26.0% ±4.4	31%	low	5 Aug 2026	Epoch ↗
FrontierMath-Tiers-1-3-v2-Private	math	51.2% ±3.0	55%	—	11 Jun 2026	Eval log ↗
FrontierMath-Tier-4-v2-Private	math	17.1% ±5.9	17%	—	11 Jun 2026	Eval log ↗
ProofBench	math	15.0%	15%	—	—	External ↗
Chess Puzzles	games	40.0% ±4.9	56%	high	6 Aug 2026	Eval log ↗
SimpleQA Verified	knowledge	66.8% ±1.5	88%	high	27 Aug 2026	Eval log ↗
FrontierMath-Tier-4-2025-07-01-Privatesuperseded	math	4.2% ±2.9	9%	—	17 Dec 2025	Epoch ↗
DeepResearch Bench	agents	49.8%	90%	low	—	External ↗
GSO-Bench	coding	9.8%	21%	—	—	External ↗
Terminal Bench	agents	64.3% ±2.8	76%	agent: Junie CLI	17 Dec 2025	External ↗
ARC-AGI-2	reasoning	33.6%	35%	—	—	External ↗
GeoBench	multimodal	88.0%	100%	—	—	External ↗
FrontierMath-2025-02-28-Privatesuperseded	math	35.6% ±2.8	68%	—	17 Dec 2025	Epoch ↗
VPCT	multimodal	72.6%	80%	—	—	External ↗
WeirdML	coding	61.6% ±0.0	66%	—	—	External ↗
OTIS Mock AIME 2024-2025	math	95.6% ±3.1	96%	high	6 Aug 2026	Eval log ↗
Balrog	games	48.1%	70%	—	—	External ↗
SimpleBench	reasoning	61.1%	75%	—	—	External ↗
SWE-Bench verified	coding	75.4% ±2.0	90%	—	18 Feb 2026	Eval log ↗
GPQA diamond	science	89.4% ±2.2	93%	high	6 Aug 2026	Eval log ↗
ARC-AGI	reasoning	84.7%	86%	—	—	External ↗

Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.

API list price over time

Price history

$ per 1M tokens

10 observations (OpenRouter listing + Internet Archive snapshots). History accumulates with every ingest run; a single point means no change has been observed yet.