Gemini 3.1 Flash Lite is Google's high-efficiency model optimized for cost-sensitive, high-volume use cases. It outperforms Gemini 2.5 Flash Lite on overall quality and approaches Gemini 2.5 Flash performance across key capabilities including audio input, RAG snippet ranking, translation, data extraction, and code completion. It supports full thinking levels (minimal/low/medium/high) for fine-grained cost-performance trade-offs, and is priced at half the cost of Gemini 3 Flash.
API|VisionReasoningWeb SearchFile|Proprietary Model
Knowledge Cutoff
2025-01-31
Input → Output Format
Context Memory
1.0MIN66KOUT
AI Performance Evaluation
Arena Overall Score
1439
±5As of 2026-04-23
Overall Rank
No.48
20,088 Votes
Arena by Ability
Hard Prompts
1448±6No.60
Expert Knowledge
1449±15No.60
Instruction Following
1411±8No.71
Conversation Memory
1447±10No.46
Creative
1419±11No.40
Coding
1461±9No.77
Math
1438±16No.42
Arena by Occupation
Creative Writing
1424±9No.42
Social Sciences
1462±10No.42
Media
1411±10No.45
Business
1433±10No.54
Healthcare
1462±16No.52
Legal
1443±15No.58
Software
1460±7No.61
Mathematics
1431±18No.62
Source:Arena Intelligence
Overall
AA Intelligence Index
34%↓5%
LiveBench
62%↑2%
Reasoning & Math
GPQA Diamond
82%↑1%
HLE
16%↓1%
LB Reasoning
60%↑0%
LB Math
74%↑0%
LB Data
55%↑5%
Coding
AA Coding Index
30%↓4%
LB Coding
69%↓5%
LB Agentic
33%↓10%
TAU2
31%↓42%
TerminalBench
24%↓7%
SciCode
42%↑1%
Language & Instructions
IFBench
77%↑20%
AA-LCR
65%↑4%
Hallucination (HHEM)
8.2%↓2%
Factual (HHEM)
92%↑2%
LB Language
73%↑1%
LB IF
69%↑23%
Output Speed
Standard Mode
335tok/s↑253
First Output 5.02s
