Inference360
Google

Gemini 3.7 Flash

Gemini 3.7 Flash is a reasoning model from Google. 51 benchmarks count toward its score, in 8 categories.

availableShows if the model has enough results for an index.
IndexOverall score. 50 is the middle.72.3 ±2.3
CoverageShare of the index weight with results.100%
SpeedOutput tokens per second.80/s
Input / 1MUS dollars per 1M input tokens.$0.75 batch $0.375
Output / 1MUS dollars per 1M output tokens.$3.75 batch $1.88 US dollars per 1M output tokens in a batch.
ContextMaximum tokens in one request.1.05M
EloLMArena rating and rank.1490 (#8)

50 is the middle of the board. The range shows the doubt in the index. Batch work costs less.

5,640 votes. Elo shows what people prefer. It does not change the score.

CapabilitiesScore per category. 50 is the middle.

50 is the middle
AgenticMulti-step tasks with tools.
70.7
CodingCode writing and repair.
70.1
ReasoningLogic problems and puzzles.
71.0
MultimodalTasks with images and text.
70.3
KnowledgeFacts and expert knowledge.
72.8
MultilingualTasks in many languages.
94.3
InstructionTasks with strict rules in the prompt.
86.0
MathMath problems.
69.8

Results

51 counted
BenchmarkThe test name.CategoryThe capability that the test measures.ResultThe score from the publisher.IndexThis result on the index scale.RunThe settings of the run.DateDate of the result.Published byThe source of the result.
OTIS Mock AIME 2024-2025Math97.2%65.9high effort—Epoch AI
OpenAI MRCR v2 8-needle 64K-128KReasoning97.0%———OpenAI
ARC-AGI-1 (semi-private)Reasoning95.5%72.3high effort—ARC Prize Foundation
GPQA diamondKnowledge94.8%65.8high effort—Epoch AI
Artificial Analysis GPQA DiamondKnowledge94.5%65.4——Artificial Analysis
GPQA DiamondKnowledge93.9%65.0—1 Sept 2026Vals AI
LiveBench MathematicsMath93.5%72.8high effort25 Jun 2026LiveBench
Artificial Analysis Harvey LAB-AAAgentic90.7%73.6——Artificial Analysis
MMLU ProKnowledge90.1%62.5—1 Sept 2026Vals AI
MMMU ProMultimodal89.0%72.1—1 Sept 2026Vals AI
CharXiv ReasoningMultimodal88.7%67.1——CharXiv authors
LiveCodeBenchCoding88.7%64.9—1 Sept 2026Vals AI
LiveBench ReasoningReasoning87.8%77.9high effort25 Jun 2026LiveBench
BioMysteryBench Human SolvableKnowledge87.1%———Anthropic
Terminal-Bench 2.1 (provider run)Agentic85.8%74.7——DeepSeek-AI
Terminal-Bench 2.1 (provider run)Agentic85.8%74.7——DeepSeek-AI
Artificial Analysis MMMU-ProMultimodal85.5%71.6——Artificial Analysis
LiveBench LanguageKnowledge85.5%73.6high effort25 Jun 2026LiveBench
LVBenchMultimodal85.4%———Qwen Team
ARC-AGI-2 (semi-private)Reasoning84.6%82.4high effort—ARC Prize Foundation
CharXiv Reasoning without toolsMultimodal84.5%———CharXiv authors
LABBench2: An Improved Benchmark for AI Systems Performing Biology ResearchKnowledge82.1%———Jon M. Laurent et al.
Artificial Analysis Long Context ReasoningReasoning81.7%64.8——Artificial Analysis
SWE-benchCoding80.8%62.4—1 Sept 2026Vals AI
LiveBench Instruction FollowingInstruction79.9%86.0high effort25 Jun 2026LiveBench
LiveBench CodingCoding78.9%68.6high effort25 Jun 2026LiveBench
Terminal-Bench 2.1Agentic77.5%69.8—21 Sept 2026Vals AI
EuroEval FrenchMultilingual77.1%95.0——EuroEval
EuroEval ItalianMultilingual76.9%95.0——EuroEval
EuroEval SwedishMultilingual76.6%95.0——EuroEval
Artificial Analysis Coding IndexCoding76.1%72.6——Artificial Analysis
EuroEval PortugueseMultilingual74.4%95.0——EuroEval
EuroEval DutchMultilingual73.1%93.5——EuroEval
FrontierMath-Tiers-1-3-v2-PrivateMath71.6%71.7high effort—Epoch AI
Vibe Code Bench v1.1Coding70.4%71.5OpenHands21 Sept 2026Vals AI
SimpleQA VerifiedKnowledge69.2%85.6high effort—Epoch AI
EuroEval SpanishMultilingual68.3%87.5——EuroEval
LiveBench Data AnalysisReasoning68.0%50.4high effort25 Jun 2026LiveBench
EuroEval PolishMultilingual67.9%87.0——EuroEval
IOICoding67.8%75.2—21 Sept 2026Vals AI
EuroEval GermanMultilingual67.3%86.3——EuroEval
SkillsBenchCoding65.9%78.8OpenHands11 Sept 2026Vals AI
DeepSWEAgentic65.3%71.2——Datacurve AI
Artificial Analysis AnalystAgentAgentic60.0%83.6——Artificial Analysis
LiveBench Agentic CodingAgentic58.3%71.6high effort25 Jun 2026LiveBench
ProofBench v1.1Math58.0%72.9—21 Sept 2026Vals AI
Artificial Analysis SciCodeCoding57.2%72.2——Artificial Analysis
Artificial Analysis Omniscience AccuracyKnowledge55.3%82.3——Artificial Analysis
HLE-VerifiedKnowledge53.6%———Weiqi Zhai et al.
OSWorld 2.0Agentic47.9%78.8——Mengqi Yuan et al.
Artificial Analysis Humanity's Last ExamKnowledge47.9%76.5——Artificial Analysis
Chess PuzzlesReasoning47.0%84.3high effort—Epoch AI
GDPval-AA normalizedAgentic43.6%69.1——Artificial Analysis
FrontierCode 1.1 MainCoding43.6%72.3——Cognition
BioMysteryBench Human DifficultKnowledge43.5%———Anthropic
Artificial Analysis Intelligence IndexKnowledge39.1%71.2——Artificial Analysis
Mystery Game PuzzlesReasoning37.0%73.3high effort—Epoch AI
FrontierMath-Tier-4-v2-PrivateMath36.6%65.5high effort—Epoch AI
Artificial Analysis Agentic IndexAgentic36.4%66.9——Artificial Analysis
Code MigrationCoding34.8%67.2—21 Sept 2026Vals AI
AutomationBenchAgentic30.4%66.0——Moonshot AI
Furniture AssemblyReasoning26.7%58.5high effort—Epoch AI
Agents' Last ExamAgentic26.3%66.0——DeepSeek-AI
FrontierSWE v2Coding20.3%65.7——Proximal
ApprenticeBench: end-to-end computer use, continual learning, and long-horizon agency on a real accounts-payable jobAgentic16.0%69.0——NeoCognition
Terminal-Bench 3.0Agentic14.9%66.2——Ryan Marten et al.
Critical Physics TasksReasoning14.3%68.4——Artificial Analysis
Terminal-Bench 4.0.0Agentic11.2%67.4high effort · mini-SWE-agent21 Sept 2026Terminal-Bench
Terminal-Bench 4.0Agentic6.1%63.7—21 Sept 2026Vals AI
Agent Arena task outcomeAgentic2.970.6high effort15 Sept 2026LMArena
ProgramBenchCoding0.0%——21 Sept 2026Vals AI
Agent Arena steerabilityAgentic-0.367.0high effort15 Sept 2026LMArena
Agent Arena command recoveryAgentic-2.165.0high effort15 Sept 2026LMArena

51 benchmarks count, from 65 of 73 results. A grey row does not count. Too few models took that benchmark.

Sources

BenchLM benchmark aggregationUsed with attribution; per-benchmark results credited to their original authorsOpenRouter, collected directlyNo licence statedEpoch AI, collected directlyCC BY — free to use and redistribute with attributionARC Prize Foundation, collected directlyNo licence stated. Their terms ask for written permission before commercial useVals AI, collected directlyNo licence stated. Read from the public leaderboard and credited to Vals AILiveBench, collected directlyNo licence stated for the leaderboard. The site repo serving these CSVs has no LICENSE; the harness repo carries upstream Apache-2.0 and MIT copies that cover code, not resultsEuroEval, collected directlyMIT — the leaderboard site and its CSV routes are in the licensed repositoryTerminal-Bench, collected directlyNo licence stated for the leaderboard. The harness repo is Apache-2.0LMArena, collected directlyCC BY 4.0 (lmarena-ai/leaderboard-dataset on Hugging Face)

Same level, lower price

DeepSeek V4.1 Flash70.6 · $0.6MiMo-V2.6-Pro73.5 · $0.87GPT-5.6 Luna69.9 · $1.2

More from Google

Gemini 3.8 Flash72.1Gemini 3.5 Flash67.3Gemini 3.6 Flash66.8Gemini 3.1 Pro64.7Gemini 3 Pro60.5