BenchAtlas

Rankings / Google DeepMind

Gemini 3.5 Flash

proprietary

released 2026-05-19

BenchAtlas Index

as of 2026-07-14
78.9
medium effort
rank #17
8 families · 4 categories · medium
78.4
high effort
rank #21
8 families · 4 categories · medium
72.2
high effort
rank #55
16 families · 5 categories · high

Benchmark evidence

106 results
agentic coding
Release 2026-06-25
tasks_counted=3 · livebench_version=2026-06-25
high effort49.0 %independentLiveBench
τ²-Benchmedium effort95.6 %independentArtificial Analysis
τ²-Benchhigh effort95.3 %independentArtificial Analysis
τ²-Benchminimal effort58.8 %independentArtificial Analysis
τ²-Bench
subset=banking
high effort25.4 %independentArtificial Analysis
coding
Release 2026-06-25
tasks_counted=2 · livebench_version=2026-06-25
high effort78.2 %independentLiveBench
SciCodehigh effort53.1 %independentArtificial Analysis
SciCodemedium effort53.0 %independentArtificial Analysis
SciCodeminimal effort48.8 %independentArtificial Analysis
data analysis
Release 2026-06-25
tasks_counted=3 · livebench_version=2026-06-25
high effort64.9 %independentLiveBench
external indices
AA Coding Indexhigh effort70.1 pointsindependentArtificial Analysis
Intelligence Index v4.1high effort50.2 pointsindependentArtificial Analysis
Intelligence Index v4.1medium effort45.4 pointsindependentArtificial Analysis
Intelligence Index v4.1minimal effort34.9 pointsindependentArtificial Analysis
ECIhigh effort154.6 pointsindependentEpoch AI Benchmarking Hub
ECI154.6 pointsindependentEpoch AI Benchmarking Hub
ECIminimal effort154.6 pointsindependentEpoch AI Benchmarking Hub
Vals Indexhigh effort62.7 pointsindependentVals AI
factuality
SimpleQA Verifiedhigh effort68.4 %independentEpoch AI Benchmarking Hub
2026-05-27
human preference
Chinese (style control)older version
arena=text · category=chinese · style_control=true
medium effort1529.8
1502.2–1557.4
communityLMArena Leaderboard Dataset
Math (style control)older version
arena=text · category=math · style_control=true
high effort1517.6
1492.2–1543.0
communityLMArena Leaderboard Dataset
Expert (style control)older version
arena=text · category=expert · style_control=true
high effort1508.4
1489.4–1527.5
communityLMArena Leaderboard Dataset
Coding (style control)older version
arena=text · category=coding · style_control=true
medium effort1507.9
1496.8–1519.0
communityLMArena Leaderboard Dataset
Industry Mathematical (style control)older version
arena=text · category=industry_mathematical · style_control=true
high effort1507.9
1483.5–1532.3
communityLMArena Leaderboard Dataset
Chinese (style control)older version
arena=text · category=chinese · style_control=true
high effort1507.7
1482.0–1533.4
communityLMArena Leaderboard Dataset
Coding (style control)older version
arena=text · category=coding · style_control=true
high effort1506.0
1494.3–1517.6
communityLMArena Leaderboard Dataset
Industry Medicine And Healthcare (style control)older version
arena=text · category=industry_medicine_and_healthcare · style_control=true
medium effort1500.8
1478.6–1522.9
communityLMArena Leaderboard Dataset
Industry Software And It Services (style control)older version
arena=text · category=industry_software_and_it_services · style_control=true
high effort1499.8
1489.9–1509.7
communityLMArena Leaderboard Dataset
Russian (style control)older version
arena=text · category=russian · style_control=true
high effort1497.4
1478.1–1516.8
communityLMArena Leaderboard Dataset
Industry Life And Physical And Social Science (style control)older version
arena=text · category=industry_life_and_physical_and_social_science · style_control=true
high effort1496.6
1481.5–1511.7
communityLMArena Leaderboard Dataset
Industry Software And It Services (style control)older version
arena=text · category=industry_software_and_it_services · style_control=true
medium effort1495.1
1485.5–1504.7
communityLMArena Leaderboard Dataset
Hard Prompts (style control)older version
arena=text · category=hard_prompts · style_control=true
medium effort1494.4
1486.4–1502.4
communityLMArena Leaderboard Dataset
Industry Life And Physical And Social Science (style control)older version
arena=text · category=industry_life_and_physical_and_social_science · style_control=true
medium effort1494.0
1479.2–1508.7
communityLMArena Leaderboard Dataset
Russian (style control)older version
arena=text · category=russian · style_control=true
medium effort1493.8
1474.9–1512.7
communityLMArena Leaderboard Dataset
Industry Medicine And Healthcare (style control)older version
arena=text · category=industry_medicine_and_healthcare · style_control=true
high effort1492.0
1469.4–1514.6
communityLMArena Leaderboard Dataset
Hard Prompts (style control)older version
arena=text · category=hard_prompts · style_control=true
high effort1491.7
1483.8–1499.7
communityLMArena Leaderboard Dataset
Expert (style control)older version
arena=text · category=expert · style_control=true
medium effort1489.6
1471.3–1507.9
communityLMArena Leaderboard Dataset
Multi Turn (style control)older version
arena=text · category=multi_turn · style_control=true
high effort1488.3
1473.7–1502.8
communityLMArena Leaderboard Dataset
Hard Prompts English (style control)older version
arena=text · category=hard_prompts_english · style_control=true
medium effort1485.3
1474.7–1496.0
communityLMArena Leaderboard Dataset
Exclude Ties (style control)older version
arena=text · category=exclude_ties · style_control=true
medium effort1484.5
1475.6–1493.5
communityLMArena Leaderboard Dataset
Exclude Ties (style control)older version
arena=text · category=exclude_ties · style_control=true
high effort1484.5
1475.9–1493.1
communityLMArena Leaderboard Dataset
Hard Prompts English (style control)older version
arena=text · category=hard_prompts_english · style_control=true
high effort1484.5
1473.6–1495.4
communityLMArena Leaderboard Dataset
Multi Turn (style control)older version
arena=text · category=multi_turn · style_control=true
medium effort1482.3
1467.8–1496.8
communityLMArena Leaderboard Dataset
Longer Query (style control)older version
arena=text · category=longer_query · style_control=true
high effort1480.1
1470.2–1490.0
communityLMArena Leaderboard Dataset
Industry Legal And Government (style control)older version
arena=text · category=industry_legal_and_government · style_control=true
medium effort1478.9
1457.6–1500.2
communityLMArena Leaderboard Dataset
English (style control)older version
arena=text · category=english · style_control=true
high effort1476.5
1467.5–1485.6
communityLMArena Leaderboard Dataset
Overall (style control)
arena=text · category=overall · style_control=true
high effort1476.5
1469.9–1483.0
communityLMArena Leaderboard Dataset
Industry Legal And Government (style control)older version
arena=text · category=industry_legal_and_government · style_control=true
high effort1475.0
1452.9–1497.0
communityLMArena Leaderboard Dataset
Overall (style control)
arena=text · category=overall · style_control=true
medium effort1474.9
1468.1–1481.7
communityLMArena Leaderboard Dataset
Longer Query (style control)older version
arena=text · category=longer_query · style_control=true
medium effort1473.3
1463.8–1482.7
communityLMArena Leaderboard Dataset
Non English (style control)older version
arena=text · category=non_english · style_control=true
medium effort1473.2
1464.6–1481.8
communityLMArena Leaderboard Dataset
Industry Writing And Literature And Language (style control)older version
arena=text · category=industry_writing_and_literature_and_language · style_control=true
medium effort1472.8
1460.6–1485.0
communityLMArena Leaderboard Dataset
English (style control)older version
arena=text · category=english · style_control=true
medium effort1472.3
1463.4–1481.3
communityLMArena Leaderboard Dataset
Non English (style control)older version
arena=text · category=non_english · style_control=true
high effort1471.7
1463.0–1480.4
communityLMArena Leaderboard Dataset
Industry Mathematical (style control)older version
arena=text · category=industry_mathematical · style_control=true
medium effort1471.3
1445.2–1497.4
communityLMArena Leaderboard Dataset
Creative Writing (style control)older version
arena=text · category=creative_writing · style_control=true
medium effort1470.0
1455.0–1484.9
communityLMArena Leaderboard Dataset
Creative Writing (style control)older version
arena=text · category=creative_writing · style_control=true
high effort1464.0
1447.7–1480.2
communityLMArena Leaderboard Dataset
Industry Writing And Literature And Language (style control)older version
arena=text · category=industry_writing_and_literature_and_language · style_control=true
high effort1462.6
1449.7–1475.5
communityLMArena Leaderboard Dataset
Instruction Following (style control)older version
arena=text · category=instruction_following · style_control=true
medium effort1462.6
1452.2–1473.0
communityLMArena Leaderboard Dataset
Instruction Following (style control)older version
arena=text · category=instruction_following · style_control=true
high effort1462.5
1451.9–1473.2
communityLMArena Leaderboard Dataset
Industry Entertainment And Sports And Media (style control)older version
arena=text · category=industry_entertainment_and_sports_and_media · style_control=true
medium effort1459.4
1446.6–1472.2
communityLMArena Leaderboard Dataset
Industry Business And Management And Financial Operations (style control)older version
arena=text · category=industry_business_and_management_and_financial_operations · style_control=true
medium effort1459.0
1445.5–1472.5
communityLMArena Leaderboard Dataset
Industry Business And Management And Financial Operations (style control)older version
arena=text · category=industry_business_and_management_and_financial_operations · style_control=true
high effort1457.4
1443.8–1471.0
communityLMArena Leaderboard Dataset
Industry Entertainment And Sports And Media (style control)older version
arena=text · category=industry_entertainment_and_sports_and_media · style_control=true
high effort1449.1
1435.1–1463.2
communityLMArena Leaderboard Dataset
knowledge science
GPQA Diamondhigh effort92.8 %independentEpoch AI Benchmarking Hub
2026-05-22
GPQA Diamond
implementation=vals-ai
high effort92.7 %independentVals AI
GPQA Diamond
implementation=artificial-analysis
high effort92.2 %independentArtificial Analysis
GPQA Diamond
implementation=artificial-analysis
medium effort92.1 %independentArtificial Analysis
GPQA Diamond
implementation=artificial-analysis
minimal effort82.8 %independentArtificial Analysis
Humanity's Last Exam
implementation=artificial-analysis
high effort41.0 %independentArtificial Analysis
Humanity's Last Exam
implementation=artificial-analysis
medium effort39.9 %independentArtificial Analysis
Humanity's Last Exam
implementation=artificial-analysis
minimal effort23.1 %independentArtificial Analysis
MMLU-Pro
implementation=vals-ai
high effort89.5 %independentVals AI
language
Release 2026-06-25
tasks_counted=3 · livebench_version=2026-06-25
high effort84.6 %independentLiveBench
long context instruction
AA-LCRmedium effort71.0 %independentArtificial Analysis
AA-LCRhigh effort69.3 %independentArtificial Analysis
AA-LCRminimal effort53.3 %independentArtificial Analysis
IFBenchhigh effort76.3 %independentArtificial Analysis
IFBenchmedium effort74.6 %independentArtificial Analysis
IFBenchminimal effort47.3 %independentArtificial Analysis
Release 2026-06-25
tasks_counted=4 · livebench_version=2026-06-25
high effort75.6 %independentLiveBench
multimodal
MMMU
implementation=vals-ai
high effort88.3 %independentVals AI
professional
CorpFinhigh effort64.7 %independentVals AI
LegalBenchhigh effort83.6 %independentVals AI
TaxEvalhigh effort74.4 %independentVals AI
reasoning math
ARC-AGI-1older version
split=public_eval · model_type=CoT
high effort96.0 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
high effort92.5 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
high effort72.1 %independentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
high effort72.1 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
minimal effort58.0 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
minimal effort48.8 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
minimal effort9.2 %independentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
minimal effort8.9 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
high effort0.9 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
high effort0.9 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
high effort0.4 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
high effort0.3 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
minimal effort0.1 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
minimal effort0.1 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
minimal effort0.1 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
minimal effort0.1 usd_per_taskindependentARC Prize Leaderboard
FrontierMathhigh effort39.0 %independentEpoch AI Benchmarking Hub
2026-05-22
FrontierMath Tier 4older versionhigh effort14.6 %independentEpoch AI Benchmarking Hub
2026-05-25
Release 2026-06-25
tasks_counted=4 · livebench_version=2026-06-25
high effort88.2 %independentLiveBench
Release 2026-06-25
tasks_counted=4 · livebench_version=2026-06-25
high effort82.0 %independentLiveBench
OTIS Mock AIME 2024–2025high effort95.6 %independentEpoch AI Benchmarking Hub
2026-05-25

Agent + model results

systems, not bare-model scores
agent + model Epoch Inspect harness + Gemini 3.5 FlashSWE-bench Verified79.3 %independentEpoch AI Benchmarking Hub
agent + model Artificial Analysis harness + Gemini 3.5 FlashTerminal-Bench 2.178.7 %independentArtificial Analysis
agent + model Artificial Analysis harness + Gemini 3.5 FlashTerminal-Bench Hard46.2 %independentArtificial Analysis
agent + model Artificial Analysis harness + Gemini 3.5 FlashTerminal-Bench Hard40.9 %independentArtificial Analysis
agent + model Artificial Analysis harness + Gemini 3.5 FlashTerminal-Bench Hard39.4 %independentArtificial Analysis

These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.