01Overview02Models03Rankings04Compare05Quiz06Guide

Model

Gemini 3.5 Flash NO.18

Fast, cheap multimodal workhorse with near-Pro quality.

CompareBack to the index

Google DeepMind · Gemini · Fast · Closed weights

Specification

Fig.01 — Gemini — Google DeepMind

USA

Context window
1.0Mtokens
Max output
66Ktokens
Input price
$1.50/ 1M
Output price
$9/ 1M
Cached input
$0.15/ 1M
Throughput
—
Class
Fast
Modalities
Text, Image, Audio, Video, Pdf
Weights
Closed
Knowledge cutoff
Jan 2025
Released
May 2026

Overview

Gemini 3.5 Flash is Google's newest fast tier, balancing strong quality with low latency and a 1M-token context. Google markets it as beating the older Pro flagship on coding and agentic work, and it is an excellent-value multimodal workhorse. As with the wider Gemini line, its real-world reliability on long autonomous coding trails the Claude/GPT frontier even where benchmarks look competitive.

Strengths

  • Excellent price-to-quality for high-volume work
  • Fast multimodal (image/video/PDF) processing
  • Full 1M-token context window
  • Strong general reasoning for its tier

Trade-offs

  • Agentic coding reliability below the frontier
  • Weaker than Pro on the hardest reasoning
  • Output capped at ~64K tokens

Fit

Best for

  • High-volume multimodal pipelines
  • Chat and summarization at scale
  • Cost-sensitive general assistants
  • Fast document and media understanding

Not ideal for

  • Unsupervised long-horizon coding agents
  • Frontier-hard reasoning problems

Capabilities

Ten axes, normalized 0–100 and scored the same way across the whole catalog — so a 78 here means what a 78 means anywhere else on the bench.

Fig.02 — Capability radar

0 — 100

Reasoning
83
Coding
78
Math
87
Writing
83
Knowledge
87
Speed
88
Agentic
75
Vision
89
Multilingual
87
Long Context
90

Benchmarks

Public results, with independent third-party runs marked. Bars normalize percentages against 100 and Elo ratings against a 1500 ceiling.

Independent index

Artificial Analysis Intelligence Index

Composite of ~9–10 independent evals · Artificial Analysis

50/ 100

BenchmarkResult
Artificial Analysis Intelligence IndexIndependentGeneral · Artificial Analysis
50%
SWE-bench VerifiedIndependentCoding · SWE-bench
79.3%
GPQA DiamondIndependentReasoning · Independent aggregators
82.8%
LiveBenchIndependentGeneral · LiveBench
75%
DeepSWEIndependentCoding · Datacurve
28%

Alternatives

Models in roughly the same class — the ones worth weighing against this record.

Compare side by side
01GPT-5.4 miniOpenAILow-cost previous-generation OpenAI workhorse for standard routing.02Claude Haiku 4.5AnthropicFastest, cheapest Claude — snappy everyday tasks at scale.03Gemini 3.1 Flash-LiteGoogle DeepMindCheapest Gemini, optimized for low-latency high-volume traffic.

Index

PreviousGemini 3.1 ProNextGemini 3.1 Flash-Lite
All models/EDUCATION