Gemini 3 Flash

Leaderboards

Where Gemini 3 Flash sits among the top 20 in each category.

By task

Scored from votes on each kind of prompt alone, on the same 1,000-point scale.

Test results

Every test Gemini 3 Flash has a published result on, out of 100%, with its rank among the models that took it.

• GPQA

  #25 / 256

  Graduate-level science questions written to be hard to look up.

  90.4%

  Verified

• AIME 2025

  #7 / 123

  Competition maths problems from the 2025 American Invitational Mathematics Exam.

  99.7%

  Self-reported

• SWE-Bench Verified

  #22 / 120

  Fixing real bugs in real GitHub projects, checked by the projects' own tests.

  78.0%

  Self-reported

• Humanity's Last Exam

  #33 / 108

  Humanity's Last Exam (HLE) is a multi-modal academic benchmark with 2,500 questions across mathematics, humanities, and natural sciences, designed to test LLM capabilities at the frontier of human knowledge with unambiguous, verifiable solutions

  43.5%

  Self-reported · In the text score· In the vision score

• MMMU-Pro

  #9 / 73

  College-level questions that need both an image and text to answer.

  81.2%

  Self-reported · In the vision score

• CharXiv-R

  #28 / 60

  Reading and reasoning about charts from scientific papers.

  80.3%

  Self-reported

• Terminal-Bench 2.0

  #38 / 54

  Getting real work done in a command-line terminal.

  47.6%

  Self-reported

• MMMLU

  #3 / 51

  Multilingual Massive Multitask Language Understanding dataset released by OpenAI, featuring professionally translated MMLU test questions across 14 languages including Arabic, Bengali, German, Spanish, French, Hindi, Indonesian, Italian, Japanese, Korean, Portuguese, Swahili, Yoruba, and Chinese. Contains approximately 15,908 multiple-choice questions per language covering 57 subjects.

  91.8%

  Self-reported

• SimpleQA

  #7 / 47

  SimpleQA is a factuality benchmark developed by OpenAI that measures the short-form factual accuracy of large language models. The benchmark contains 4,326 short, fact-seeking questions that are adversarially collected and designed to have single, indisputable answers. Questions cover diverse topics from science and technology to entertainment, and the benchmark also measures model calibration by evaluating whether models know what they know.

  68.7%

  Self-reported

• Toolathlon

  #24 / 42

  Tool Decathlon is a comprehensive benchmark for evaluating AI agents' ability to use multiple tools across diverse task categories. It measures proficiency in tool selection, sequencing, and execution across ten different tool-use scenarios.

  49.4%

  Self-reported · In the agents score

• LiveBench

  #24 / 38

  LiveBench is a challenging, contamination-limited LLM benchmark that addresses test set contamination by releasing new questions monthly based on recently-released datasets, arXiv papers, news articles, and IMDb movie synopses. It comprises tasks across math, coding, reasoning, language, instruction following, and data analysis with verifiable, objective ground-truth answers.

  72.4%

  Self-reported · In the text score

• MCP Atlas

  #35 / 38

  MCP Atlas is a benchmark for evaluating AI models on scaled tool use capabilities, measuring how well models can coordinate and utilize multiple tools across complex multi-step tasks.

  57.4%

  Self-reported

• MRCR v2 (8-needle)

  #25 / 32

  MRCR v2 (8-needle) is a variant of the Multi-Round Coreference Resolution benchmark that includes 8 needle items to retrieve from long contexts. This tests models' ability to simultaneously track and reason about multiple pieces of information across extended conversations.

  22.1%

  Self-reported · In the text score

• Finance Agent v2

  #16 / 28

  Finance Agent v2 is an agentic financial-analysis benchmark from Vals that evaluates models on real-world finance workflows, measuring their ability to retrieve and reason over financial documents, perform multi-step calculations, and produce accurate analyses.

  42.5%

  Self-reported

• VideoMMMU

  #2 / 28

  Video-MMMU evaluates Large Multimodal Models' ability to acquire knowledge from expert-level professional videos across six disciplines through three cognitive stages: perception, comprehension, and adaptation. Contains 300 videos and 900 human-annotated questions spanning Art, Business, Science, Medicine, Humanities, and Engineering.

  86.9%

  Self-reported

• ScreenSpot Pro

  #12 / 27

  ScreenSpot-Pro is a novel GUI grounding benchmark designed to rigorously evaluate the grounding capabilities of multimodal large language models (MLLMs) in professional high-resolution computing environments. The benchmark comprises 1,581 instructions across 23 applications spanning 5 industries and 3 operating systems, featuring authentic high-resolution images from professional domains with expert annotations. Unlike previous benchmarks that focus on cropped screenshots in consumer applications, ScreenSpot-Pro addresses the complexity and diversity of real-world professional software scenarios, revealing significant performance gaps in current MLLM GUI perception capabilities.

  69.1%

  Self-reported

• HealthBench Hard

  #14 / 24

  A challenging variation of HealthBench that evaluates large language models' performance and safety in healthcare through 5,000 multi-turn conversations with particularly rigorous evaluation criteria validated by 262 physicians from 60 countries

  22.9%

  Verified

• t2-bench

  #2 / 23

  t2-bench is a benchmark for evaluating agentic tool use capabilities, measuring how well models can select, sequence, and utilize tools to solve complex tasks. It tests autonomous planning and execution in multi-step scenarios.

  90.2%

  Self-reported

• FACTS Grounding

  #16 / 20

  A benchmark evaluating language models' ability to generate factually accurate and well-grounded responses based on long-form input context, comprising 1,719 examples with documents up to 32k tokens requiring detailed responses that are fully grounded in provided documents

  61.9%

  Self-reported

• OmniDocBench 1.5

  #17 / 20

  OmniDocBench 1.5 is a comprehensive benchmark for evaluating multimodal large language models on document understanding tasks, including OCR, document parsing, information extraction, and visual question answering across diverse document types. Lower Overall Edit Distance scores are better.

  12.1%

  Self-reported

• ARC-AGI v2

  #13 / 19

  ARC-AGI-2 is an upgraded benchmark for measuring abstract reasoning and problem-solving abilities in AI systems through visual grid transformation tasks. It evaluates fluid intelligence via input-output grid pairs (1x1 to 30x30) using colored cells (0-9), requiring models to identify underlying transformation rules from demonstration examples and apply them to test cases. Designed to be easy for humans but challenging for AI, focusing on core cognitive abilities like spatial reasoning, pattern recognition, and compositional generalization.

  33.6%

  Self-reported · In the text score· In the vision score

• Global PIQA

  #2 / 15

  Global PIQA is a multilingual commonsense reasoning benchmark that evaluates physical interaction knowledge across 100 languages and cultures. It tests AI systems' understanding of physical world knowledge in diverse cultural contexts through multiple choice questions about everyday situations requiring physical commonsense.

  92.8%

  Self-reported

• Legal Agent Benchmark

  #12 / 15

  The Legal Agent Benchmark (LAB) is Harvey's open-source benchmark for evaluating AI agents on complex, long-horizon legal work. Tasks are scored under an all-pass standard against expert-curated rubrics, where a task passes only if every required rubric criterion (facts, conclusions, citations, structure, and analytical moves) passes.

  0.0%

  Self-reported

• HealthBench (hard)

  #6 / 13

  22.9%

  Unverified

• GPQA (diamond)

  #6 / 10

  Graduate-level Google-Proof Q&A benchmark with paper-aligned full and diamond subsets.

  82.3%

  Unverified

• LiveCodeBench Pro

  #3 / 5

  LiveCodeBench Pro is an advanced evaluation benchmark for large language models for code that uses Elo ratings to rank models based on their performance on coding tasks. It evaluates models on real-world coding problems from programming contests (LeetCode, AtCoder, CodeForces) and provides a relative ranking system where higher Elo scores indicate superior performance.

  77.2%

  Self-reported

• Vending-Bench 2

  #4 / 4

  Vending-Bench 2 tests longer horizon planning capabilities by evaluating how well AI models can manage a simulated vending machine business over extended periods. The benchmark measures a model's ability to maintain consistent tool usage and decision-making for a full simulated year of operation, driving higher returns without drifting off task.

  363500.0%

  Self-reported