Back to the index
Google DeepMind · Gemini
No. 15

Gemini 3.1 Flash-Lite

Cheapest Gemini, optimized for low-latency high-volume traffic.

FastClosed weightsText, Image, Audio, Video, Pdf
Field notes
Context window
1.0Mtokens
Max output
66Ktokens
Input price
$0.25/ 1M tokens
Output price
$1.50/ 1M tokens
Cached input
$0.05/ 1M tokens
Modalities
Text, Image, Audio, Video, Pdf
Knowledge cutoff
Jan 2025
Released
May 2026
Approx. speed
—
Overview

What this model is

Gemini 3.1 Flash-Lite is Google's most cost-efficient Gemini tier, tuned for ultra-low-latency, high-volume workloads. It keeps a full 1M-token multimodal context and posts surprisingly strong scores for its class, but it is optimized for cost and speed, not reasoning depth. It is the default choice when throughput, latency, and cost matter more than intelligence.

Strengths

  • Extremely low price per token
  • Very high throughput and low latency
  • Full 1M-token multimodal context
  • Solid quality for simple, high-volume tasks

Trade-offs

  • Weak on hard reasoning and complex coding
  • Not suited to agentic autonomy
  • Trails Flash and Pro on complex tasks

Best for

  • High-volume classification and extraction
  • Cheap summarization and chat
  • Latency-sensitive consumer features
  • Routing and pre-processing pipelines
  • Bulk multimodal tagging

Not ideal for

  • Frontier reasoning or research
  • Any autonomous agent workflow
Capabilities

Capability profile

Normalized 0–100 scores, comparable across the whole catalog.

Reasoning70Coding62Math72Writing75Knowledge78Speed96Agentic55Vision82Multilingual82Long Context86
Reasoning
70
Coding
62
Math
72
Writing
75
Knowledge
78
Speed
96
Agentic
55
Vision
82
Multilingual
82
Long Context
86
Benchmarks

How it scores

Public benchmark results, with independent third-party results where available. Bars normalize percentages to 100 and Elo ratings to a 1500 ceiling.

Independent index
Artificial Analysis Intelligence Index
Composite of ~9–10 independent evals · Artificial Analysis
25/ 100
BenchmarkResult
Artificial Analysis Intelligence Indexindependent
GeneralArtificial Analysis
25%
GPQA Diamondindependent
ReasoningIndependent aggregators
86.9%
MMLU-Pro
KnowledgeVendor-reported
83%
MMMU
VisionVendor-reported
76.8%
LMArena Eloindependent
GeneralLMArena
1432Elo
Alternatives

Comparable models

Models in a roughly similar class — worth weighing against this one.

GPT-5.4 nanoOpenAI

OpenAI's cheapest small model for high-volume simple work.

Claude Haiku 4.5Anthropic

Fastest, cheapest Claude — snappy everyday tasks at scale.