BenchAtlas

Rankings / Google DeepMind

Gemini 2.5 Flash

released 2025-06-17

BenchAtlas Index

as of 2026-09-11
41.4
thinking
rank #303
11 families · 5 categories · high
29.6
no reasoning
rank #381
11 families · 5 categories · high

Benchmark evidence

58 results
agentic coding
τ²-Benchthinking31.6 %independentArtificial Analysis
τ²-Benchno reasoning14.9 %independentArtificial Analysis
coding
LiveCodeBench v6
implementation=artificial-analysis
thinking69.5 %independentArtificial Analysis
LiveCodeBench v6
implementation=artificial-analysis
no reasoning49.5 %independentArtificial Analysis
SciCodethinking39.4 %independentArtificial Analysis
SciCodeno reasoning29.1 %independentArtificial Analysis
external indices
Intelligence Index v4.1thinking13.1 pointsindependentArtificial Analysis
Intelligence Index v4.1no reasoning9.9 pointsindependentArtificial Analysis
AA Math Indexthinking73.3 pointsindependentArtificial Analysis
AA Math Indexno reasoning60.3 pointsindependentArtificial Analysis
ECI140.6 pointsindependentEpoch AI Benchmarking Hub
human preference
Chinese (style control)older version
arena=text · category=chinese · style_control=true
1438.6
1431.7–1445.5
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
1430.5
1425.7–1435.2
communityLMArena Leaderboard Dataset
Industry Medicine And Healthcare (style control)older version
arena=text · category=industry_medicine_and_healthcare · style_control=true
1426.8
1419.5–1434.1
communityLMArena Leaderboard Dataset
Industry Legal And Government (style control)older version
arena=text · category=industry_legal_and_government · style_control=true
1426.6
1419.9–1433.4
communityLMArena Leaderboard Dataset
French (style control)older version
arena=text · category=french · style_control=true
1424.6
1411.5–1437.7
communityLMArena Leaderboard Dataset
Coding (style control)older version
arena=text · category=coding · style_control=true
1423.8
1419.5–1428.2
communityLMArena Leaderboard Dataset
Expert (style control)older version
arena=text · category=expert · style_control=true
1423.5
1416.5–1430.4
communityLMArena Leaderboard Dataset
Industry Software And It Services (style control)older version
arena=text · category=industry_software_and_it_services · style_control=true
1421.2
1417.5–1424.8
communityLMArena Leaderboard Dataset
Hard Prompts English (style control)older version
arena=text · category=hard_prompts_english · style_control=true
1421.2
1417.0–1425.3
communityLMArena Leaderboard Dataset
Hard Prompts (style control)older version
arena=text · category=hard_prompts · style_control=true
1419.6
1416.4–1422.7
communityLMArena Leaderboard Dataset
Longer Query (style control)older version
arena=text · category=longer_query · style_control=true
1417.8
1413.8–1421.8
communityLMArena Leaderboard Dataset
Industry Mathematical (style control)older version
arena=text · category=industry_mathematical · style_control=true
1416.4
1409.0–1423.8
communityLMArena Leaderboard Dataset
English (style control)older version
arena=text · category=english · style_control=true
1413.2
1409.9–1416.5
communityLMArena Leaderboard Dataset
Russian (style control)older version
arena=text · category=russian · style_control=true
1409.9
1404.2–1415.6
communityLMArena Leaderboard Dataset
Overall (style control)
arena=text · category=overall · style_control=true
1409.8
1407.3–1412.3
communityLMArena Leaderboard Dataset
German (style control)older version
arena=text · category=german · style_control=true
1406.5
1394.2–1418.7
communityLMArena Leaderboard Dataset
Math (style control)older version
arena=text · category=math · style_control=true
1405.8
1399.0–1412.6
communityLMArena Leaderboard Dataset
Spanish (style control)older version
arena=text · category=spanish · style_control=true
1405.7
1394.3–1417.0
communityLMArena Leaderboard Dataset
Multi Turn (style control)older version
arena=text · category=multi_turn · style_control=true
1403.5
1398.8–1408.1
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
1403.0
1398.8–1407.2
communityLMArena Leaderboard Dataset
Instruction Following (style control)older version
arena=text · category=instruction_following · style_control=true
1400.4
1396.5–1404.3
communityLMArena Leaderboard Dataset
Non English (style control)older version
arena=text · category=non_english · style_control=true
1400.3
1397.2–1403.3
communityLMArena Leaderboard Dataset
Polish (style control)older version
arena=text · category=polish · style_control=true
1399.3
1390.8–1407.9
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
1399.0
1394.4–1403.6
communityLMArena Leaderboard Dataset
Creative Writing (style control)older version
arena=text · category=creative_writing · style_control=true
1396.0
1390.9–1401.0
communityLMArena Leaderboard Dataset
Exclude Ties (style control)older version
arena=text · category=exclude_ties · style_control=true
1395.0
1391.6–1398.5
communityLMArena Leaderboard Dataset
Japanese (style control)older version
arena=text · category=japanese · style_control=true
1392.2
1377.9–1406.5
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
1386.9
1382.4–1391.4
communityLMArena Leaderboard Dataset
Korean (style control)older version
arena=text · category=korean · style_control=true
1372.2
1358.6–1385.8
communityLMArena Leaderboard Dataset
knowledge science
GPQA Diamond
implementation=artificial-analysis
thinking79.0 %independentArtificial Analysis
GPQA Diamond
implementation=artificial-analysis
no reasoning68.3 %independentArtificial Analysis
Humanity's Last Exam
implementation=artificial-analysis
thinking12.1 %independentArtificial Analysis
Humanity's Last Exam
implementation=artificial-analysis
no reasoning4.7 %independentArtificial Analysis
MMLU-Pro
implementation=artificial-analysis
thinking83.2 %independentArtificial Analysis
MMLU-Pro
implementation=artificial-analysis
no reasoning80.9 %independentArtificial Analysis
long context instruction
AA-LCRthinking65.3 %independentArtificial Analysis
AA-LCRno reasoning49.9 %independentArtificial Analysis
IFBenchthinking50.3 %independentArtificial Analysis
IFBenchno reasoning39.0 %independentArtificial Analysis
reasoning math
AIME
implementation=artificial-analysis
thinking82.3 %independentArtificial Analysis
AIME
year=2025 · implementation=artificial-analysis
thinking73.3 %independentArtificial Analysis
AIME
year=2025 · implementation=artificial-analysis
no reasoning60.3 %independentArtificial Analysis
AIME
implementation=artificial-analysis
no reasoning50.0 %independentArtificial Analysis
FrontierMath4.8 %independentEpoch AI Benchmarking Hub
2025-12-18
FrontierMath Tier 4older version4.2 %independentEpoch AI Benchmarking Hub
2025-12-18
MATH-500
implementation=artificial-analysis
thinking98.1 %independentArtificial Analysis
MATH-500
implementation=artificial-analysis
no reasoning93.2 %independentArtificial Analysis

Agent + model results

systems, not bare-model scores
agent + model mini-SWE-agent + Gemini 2.5 FlashSWE-bench bash-only28.7 %communitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 FlashSWE-bench Verified28.7 %communitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 FlashSWE-bench Verified0.1 usd_per_taskcommunitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 FlashSWE-bench bash-only0.1 usd_per_taskcommunitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 FlashTerminal-Bench 2.017.1 %communityTerminal-Bench Leaderboard
agent + model terminus-2 + Gemini 2.5 FlashTerminal-Bench 2.016.9 %communityTerminal-Bench Leaderboard
agent + model terminus-1 + Gemini 2.5 FlashTerminal-Bench 1.016.8 %communityTerminal-Bench Leaderboard
agent + model OpenHands + Gemini 2.5 FlashTerminal-Bench 2.016.4 %communityTerminal-Bench Leaderboard
agent + model gemini-cli + Gemini 2.5 FlashTerminal-Bench 2.015.4 %communityTerminal-Bench Leaderboard
agent + model Artificial Analysis harness + Gemini 2.5 FlashTerminal-Bench Hard13.6 %independentArtificial Analysis
agent + model Artificial Analysis harness + Gemini 2.5 FlashTerminal-Bench Hard12.1 %independentArtificial Analysis

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

Related variants

Gemini 2.5 Flash 04.17Gemini 2.5 Flash 05.20