Google DeepMindproprietary
Released: 2026-06-01

Google: Gemini 2.5 Flash Lite

Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It offers improved throughput, faster token generation, and better performance...

Intelligence Index
81.0/ 100
Calibrated multi-domain composite
Throughput Speed
130tok/sec
Output streaming throughput
Time To First Token
280ms
Initial chunk server latency
LiveBench Contam-Free
71.3%
September 2026 suite score

LiveBench Multi-Domain Evaluation Breakdown

Monthly refreshed contamination-free test sets
Logical Reasoning74.5%
Coding & Repo Repair72.9%
Mathematics (AIME)72.1%
Data Analysis & Tables68.9%
Language & Comprehension70.5%
Instruction Following71.3%

Sourced Benchmark Evaluations (61)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-3Active51.1%
93
2026-07-06independentSource ↗
ARC-AGI-2Nearing Saturation61.3%
75
2026-06-29independentSource ↗
FrontierMathActive21.1%
98
2026-06-26independentSource ↗
PutnamBenchActive39.2%
90
2026-06-25independentSource ↗
SWE-LancerDeprecated52.4%
20
2026-06-23independentSource ↗
Omni-MATHDeprecated55.7%
10
2026-06-23independentSource ↗
SoohakActive57.2%
98
2026-06-21independentSource ↗
FrontierCodeActive44.7%
98
2026-06-21independentSource ↗
Humanity's Last ExamActive34.6%
93
2026-06-19independentSource ↗
StructFlowBenchNearing Saturation60.6%
75
2026-06-19independentSource ↗
AIMENearing Saturation71.5%
75
2026-06-19independentSource ↗
OlympiadBenchActive61.7%
80
2026-06-18independentSource ↗
MathArenaActive60.2%
99
2026-06-17independentSource ↗
FireBenchActive59.9%
95
2026-06-17independentSource ↗
Terminal-BenchActive53.9%
80
2026-06-17independentSource ↗
OpenAI-MRCRActive68%
80
2026-06-16independentSource ↗
SimpleQAActive42.1%
90
2026-06-16independentSource ↗
CodeContestsDeprecated50.1%
10
2026-06-16independentSource ↗
ComplexBenchNearing Saturation58.5%
80
2026-06-15independentSource ↗
SWE-bench ProActive47.7%
90
2026-06-15independentSource ↗
LongBench-ProActive66.4%
80
2026-06-15independentSource ↗
ZeroSCROLLSNearing Saturation64.8%
78
2026-06-14independentSource ↗
MultiPL-EDeprecated70.8%
10
2026-06-14independentSource ↗
LiveCodeBenchSaturated63.1%
39
2026-06-13independentSource ↗
AgentIFNearing Saturation48.5%
80
2026-06-13independentSource ↗
MATHSaturated73.8%
25
2026-06-13independentSource ↗
InfiniteBenchNearing Saturation68.9%
65
2026-06-13independentSource ↗
MercuryDeprecated67.8%
10
2026-06-12independentSource ↗
SCROLLSNearing Saturation64.8%
65
2026-06-12independentSource ↗
MMLU-ProNearing Saturation74.5%
75
2026-06-12independentSource ↗
RULERNearing Saturation74.5%
84
2026-06-11independentSource ↗
TabMWPSaturated69.3%
25
2026-06-11independentSource ↗
CyberSecEvalDeprecated69.3%
10
2026-06-11independentSource ↗
SWE-bench VerifiedSaturated60.1%
25
2026-06-10independentSource ↗
MMLU-ReduxSaturated79.4%
25
2026-06-10independentSource ↗
QuALITYNearing Saturation71.3%
78
2026-06-10independentSource ↗
LongBenchNearing Saturation69.7%
65
2026-06-09independentSource ↗
SWE-benchDeprecated52.4%
10
2026-06-09independentSource ↗
MGSMNearing Saturation71.5%
65
2026-06-09independentSource ↗
AGIEvalSaturated67.6%
25
2026-06-08independentSource ↗
QASPERNearing Saturation67.2%
65
2026-06-08independentSource ↗
TruthfulQASaturated71.3%
25
2026-06-08independentSource ↗
GSM-HardDeprecated72.3%
10
2026-06-07independentSource ↗
SuperGLUESaturated80.2%
25
2026-06-07independentSource ↗
CodeXGLUEDeprecated72.4%
23
2026-06-06independentSource ↗
MT-BenchNearing Saturation7.3%
75
2026-06-06independentSource ↗
BIG-Bench HardSaturated77%
25
2026-06-06independentSource ↗
WinoGrandeSaturated81.7%
28
2026-06-05independentSource ↗
MBPP+Nearing Saturation78.5%
65
2026-06-05independentSource ↗
IFEvalDeprecated69.9%
25
2026-06-05independentSource ↗
GPQANearing Saturation71.3%
78
2026-06-05independentSource ↗
Needle-in-a-HaystackSaturated96.2%
44
2026-06-04independentSource ↗
HumanEval+Saturated83.2%
25
2026-06-04independentSource ↗
CommonsenseQASaturated82.5%
28
2026-06-04independentSource ↗
MBPPSaturated81.6%
25
2026-06-03independentSource ↗
GSM8KSaturated81.3%
25
2026-06-03independentSource ↗
GLUESaturated84.9%
28
2026-06-03independentSource ↗
MMLUSaturated82.6%
25
2026-06-02independentSource ↗
HumanEvalSaturated87.8%
25
2026-06-02independentSource ↗
ARC-AGISaturated73.9%
25
2026-06-02independentSource ↗
LiveBench (Continuous)Active71.3%
94
2025-02-01independentSource ↗
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI

client = OpenAI()
response = client.chat.completions.create(
    model="google-gemini-2-5-flash-lite",
    messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
    temperature=0.2
)
print(response.choices[0].message.content)

Evaluation & Sourcing Notes

Scores listed for Google: Gemini 2.5 Flash Lite represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.