BenchAtlas

Rankings / Google DeepMind

Gemini 2.5 Pro (2025-06-17)

released 2025-06-17

BenchAtlas Index

as of 2026-09-11
54.4
base configuration
rank #205
15 families · 5 categories · high

Benchmark evidence

89 results
agentic coding
τ²-Bench54.1 %independentArtificial Analysis
τ²-Bench
subset=banking
9.7 %independentArtificial Analysis
coding
LiveCodeBench v6
implementation=artificial-analysis
80.1 %independentArtificial Analysis
LiveCodeBench v5older version
problems=279 · window_start=2024-08-01
69.2 %independentLiveCodeBench Leaderboard
SciCode46.3 %independentArtificial Analysis
external indices
AA Coding Index33.3 pointsindependentArtificial Analysis
Intelligence Index v4.116.7 pointsindependentArtificial Analysis
AA Math Index87.7 pointsindependentArtificial Analysis
ECI145.3 pointsindependentEpoch AI Benchmarking Hub
ECI32K budget145.3 pointsindependentEpoch AI Benchmarking Hub
ECI16K budget145.3 pointsindependentEpoch AI Benchmarking Hub
ECI8K budget145.3 pointsindependentEpoch AI Benchmarking Hub
factuality
SimpleQA Verified56.0 %independentEpoch AI Benchmarking Hub
2025-12-08
human preference
Chinese (style control)older version
arena=text · category=chinese · style_control=true
1489.1
1482.1–1496.1
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
1471.4
1466.7–1476.2
communityLMArena Leaderboard Dataset
German (style control)older version
arena=text · category=german · style_control=true
1470.3
1457.4–1483.1
communityLMArena Leaderboard Dataset
Industry Legal And Government (style control)older version
arena=text · category=industry_legal_and_government · style_control=true
1468.8
1461.9–1475.7
communityLMArena Leaderboard Dataset
Industry Medicine And Healthcare (style control)older version
arena=text · category=industry_medicine_and_healthcare · style_control=true
1467.7
1460.3–1475.1
communityLMArena Leaderboard Dataset
Coding (style control)older version
arena=text · category=coding · style_control=true
1465.3
1461.0–1469.6
communityLMArena Leaderboard Dataset
French (style control)older version
arena=text · category=french · style_control=true
1465.0
1451.3–1478.8
communityLMArena Leaderboard Dataset
Expert (style control)older version
arena=text · category=expert · style_control=true
1460.5
1453.5–1467.5
communityLMArena Leaderboard Dataset
Industry Software And It Services (style control)older version
arena=text · category=industry_software_and_it_services · style_control=true
1459.6
1456.0–1463.3
communityLMArena Leaderboard Dataset
Hard Prompts English (style control)older version
arena=text · category=hard_prompts_english · style_control=true
1458.8
1454.7–1463.0
communityLMArena Leaderboard Dataset
Hard Prompts (style control)older version
arena=text · category=hard_prompts · style_control=true
1458.2
1455.0–1461.3
communityLMArena Leaderboard Dataset
Polish (style control)older version
arena=text · category=polish · style_control=true
1456.7
1448.1–1465.3
communityLMArena Leaderboard Dataset
Longer Query (style control)older version
arena=text · category=longer_query · style_control=true
1456.1
1452.1–1460.1
communityLMArena Leaderboard Dataset
Spanish (style control)older version
arena=text · category=spanish · style_control=true
1452.9
1441.6–1464.3
communityLMArena Leaderboard Dataset
Russian (style control)older version
arena=text · category=russian · style_control=true
1451.0
1445.2–1456.9
communityLMArena Leaderboard Dataset
English (style control)older version
arena=text · category=english · style_control=true
1449.4
1446.0–1452.8
communityLMArena Leaderboard Dataset
Multi Turn (style control)older version
arena=text · category=multi_turn · style_control=true
1448.6
1443.8–1453.4
communityLMArena Leaderboard Dataset
Industry Mathematical (style control)older version
arena=text · category=industry_mathematical · style_control=true
1448.6
1441.0–1456.2
communityLMArena Leaderboard Dataset
Japanese (style control)older version
arena=text · category=japanese · style_control=true
1448.2
1433.3–1463.2
communityLMArena Leaderboard Dataset
Overall (style control)
arena=text · category=overall · style_control=true
1445.6
1443.1–1448.1
communityLMArena Leaderboard Dataset
Exclude Ties (style control)older version
arena=text · category=exclude_ties · style_control=true
1445.3
1441.8–1448.8
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
1444.3
1440.0–1448.6
communityLMArena Leaderboard Dataset
Creative Writing (style control)older version
arena=text · category=creative_writing · style_control=true
1442.7
1437.5–1447.9
communityLMArena Leaderboard Dataset
Math (style control)older version
arena=text · category=math · style_control=true
1441.6
1434.5–1448.6
communityLMArena Leaderboard Dataset
Instruction Following (style control)older version
arena=text · category=instruction_following · style_control=true
1438.5
1434.6–1442.4
communityLMArena Leaderboard Dataset
Non English (style control)older version
arena=text · category=non_english · style_control=true
1436.9
1433.8–1440.0
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
1436.8
1432.2–1441.5
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
1429.5
1424.9–1434.2
communityLMArena Leaderboard Dataset
Korean (style control)older version
arena=text · category=korean · style_control=true
1414.1
1400.8–1427.3
communityLMArena Leaderboard Dataset
knowledge science
GPQA Diamond85.3 %independentEpoch AI Benchmarking Hub
2025-11-16
GPQA Diamond
implementation=artificial-analysis
84.4 %independentArtificial Analysis
Humanity's Last Exam
implementation=artificial-analysis
22.5 %independentArtificial Analysis
MMLU-Pro
implementation=artificial-analysis
86.2 %independentArtificial Analysis
long context instruction
AA-LCR69.0 %independentArtificial Analysis
IFBench48.7 %independentArtificial Analysis
MultiChallenge53.6 %
52.3–54.9
independentScale Labs
professional
CaseLaw63.9 %independentVals AI
CorpFin60.8 %independentVals AI
reasoning math
AIME
implementation=artificial-analysis
88.7 %independentArtificial Analysis
AIME
year=2025 · implementation=artificial-analysis
87.7 %independentArtificial Analysis
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 16K budget56.4 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 32K budget55.9 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 8K budget44.2 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 16K budget41.0 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 32K budget37.0 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 8K budget29.5 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 1K budget17.5 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 1K budget16.0 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 16K budget5.1 %independentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 32K budget4.9 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 32K budget4.6 %independentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 16K budget4.0 %independentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 8K budget2.9 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 8K budget2.6 %independentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 32K budget0.8 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 32K budget0.8 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 16K budget0.7 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 16K budget0.7 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 32K budget0.5 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 16K budget0.5 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 8K budget0.4 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 8K budget0.4 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 1K budget0.4 %independentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 32K budget0.4 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 16K budget0.4 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 8K budget0.3 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 8K budget0.3 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 1K budget0.1 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=public_eval · model_type=CoT
thinking · 1K budget0.1 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=semi_private · model_type=CoT
thinking · 1K budget0.1 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-1older version
split=public_eval · model_type=CoT
thinking · 1K budget0.0 usd_per_taskindependentARC Prize Leaderboard
ARC-AGI-2
split=semi_private · model_type=CoT
thinking · 1K budget0.0 %independentARC Prize Leaderboard
FrontierMath14.1 %independentEpoch AI Benchmarking Hub
2025-11-24
FrontierMath Tier 4older version4.2 %independentEpoch AI Benchmarking Hub
2025-07-03
MATH-500
implementation=artificial-analysis
96.7 %independentArtificial Analysis
OTIS Mock AIME 2024–202584.2 %independentEpoch AI Benchmarking Hub
2025-11-16

Agent + model results

systems, not bare-model scores
agent + model Epoch Inspect harness + Gemini 2.5 Pro (2025-06-17)SWE-bench Verified57.6 %independentEpoch AI Benchmarking Hub
agent + model mini-SWE-agent + Gemini 2.5 Pro (2025-06-17)SWE-bench Verified53.6 %communitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 Pro (2025-06-17)SWE-bench bash-only53.6 %communitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 Pro (2025-06-17)SWE-bench bash-only0.3 usd_per_taskcommunitySWE-bench Leaderboard
agent + model mini-SWE-agent + Gemini 2.5 Pro (2025-06-17)SWE-bench Verified0.3 usd_per_taskcommunitySWE-bench Leaderboard
agent + model terminus-2 + Gemini 2.5 Pro (2025-06-17)Terminal-Bench 2.032.6 %communityTerminal-Bench Leaderboard
agent + model Artificial Analysis harness + Gemini 2.5 Pro (2025-06-17)Terminal-Bench 2.128.5 %independentArtificial Analysis
agent + model Artificial Analysis harness + Gemini 2.5 Pro (2025-06-17)Terminal-Bench Hard26.5 %independentArtificial Analysis
agent + model mini-SWE-agent + Gemini 2.5 Pro (2025-06-17)Terminal-Bench 2.026.1 %communityTerminal-Bench Leaderboard
agent + model terminus-1 + Gemini 2.5 Pro (2025-06-17)Terminal-Bench 1.025.3 %communityTerminal-Bench Leaderboard
agent + model gemini-cli + Gemini 2.5 Pro (2025-06-17)Terminal-Bench 2.019.6 %communityTerminal-Bench Leaderboard
agent + model OpenHands + Gemini 2.5 Pro (2025-06-17)Terminal-Bench 2.016.4 %communityTerminal-Bench Leaderboard

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 Pro 06.05Gemini 2.5 Pro 05.06