What's Gemini 3 Flash Preview's go-to globalization or self-sufficiency?
100 AI models, 100 simple questions, asked 3 times each. Explore all 30,000 answers.
Explore by prompt
Explore by model
Gemini 3 Flash Preview
| Prompt | Run 1 | Run 2 | Run 3 | ConsensusBench |
|---|---|---|---|---|
| General | ||||
| Color | Azure | Cerulean | Azure | 0% 0 / 3 |
| Superpower | Teleportation | Teleportation | Teleportation | 100% 3 / 3 |
| Country to live in | Switzerland | Switzerland | Switzerland | 0% 0 / 3 |
| Holiday destination | Kyoto | Kyoto | Kyoto | 0% 0 / 3 |
| Roll a dice | 4 | 4 | 4 | 100% 3 / 3 |
| Sport | Tennis | Tennis | Tennis | 0% 0 / 3 |
| Day of the week | Tuesday | Wednesday | Wednesday | 66.7% 2 / 3 |
| Month of the year | October | October | October | 0% 0 / 3 |
| Number | 42 | 42 | 42 | 100% 3 / 3 |
| Letter | Q | A | Q | 33.3% 1 / 3 |
| Season | Autumn | Autumn | Autumn | 0% 0 / 3 |
| Animal | Elephant | Capybara | Wolf | 33.3% 1 / 3 |
| Fruit | Mango | Apple | Mango | 33.3% 1 / 3 |
| Vegetable | Broccoli | Carrot | Broccoli | 33.3% 1 / 3 |
| Musical instrument | Cello | Cello | Cello | 0% 0 / 3 |
| Draw a card | Eight of Spades | 7 of Diamonds | Jack of Diamonds | 0% 0 / 3 |
| Flip a coin | Heads | Heads | Heads | 100% 3 / 3 |
| Language to learn | Spanish | Mandarin | Mandarin | 33.3% 1 / 3 |
| Hobby | Gardening | Gardening | Gardening | 0% 0 / 3 |
| Transport | Bicycle | Bicycle | Bicycle | 100% 3 / 3 |
| Names | ||||
| Teacher's name | Bernier | Middleton | Maxwell | 0% 0 / 3 |
| Boy's name | Arthur | Julian | Julian | 0% 0 / 3 |
| Girl's name | Lyra | Eleanor | Seraphina | 0% 0 / 3 |
| CEO's name | Alistair Thorne | Alistair Vance | Alistair Thorne | 0% 0 / 3 |
| Nurse's name | Florence | Florence | Florence | 0% 0 / 3 |
| Scientist's name | Galileo | Galileo | Curie | 0% 0 / 3 |
| Doctor's name | Julian Thorne | Vance | Alistair Sterling | 0% 0 / 3 |
| Engineer's name | Tesla | Isambard | Ada | 0% 0 / 3 |
| Software developer's name | Ada | Ada | Ada | 0% 0 / 3 |
| Lawyer's name | Atticus Thorne | Atticus Finch | Atticus Finch | 0% 0 / 3 |
| Police officer's name | Miller | Miller | Murdoch | 0% 0 / 3 |
| Firefighter's name | Blaze | Blaze | Blaze | 100% 3 / 3 |
| Politician's name | Cicero | Cicero | Atticus Finch | 0% 0 / 3 |
| Artist's name | Kaelith | Vespera | Vespera | 0% 0 / 3 |
| Chef's name | Auguste | Escoffier | Auguste | 0% 0 / 3 |
| Farmer's name | Silas | Silas | Silas | 0% 0 / 3 |
| Journalist's name | Elias Thorne | Elias Thorne | Elias Thorne | 0% 0 / 3 |
| Athlete's name | Usain Bolt | Usain Bolt | Serena Williams | 66.7% 2 / 3 |
| Caregiver's name | Elena | Grace | Martha | 0% 0 / 3 |
| Judge's name | Justice Thorne | Solomon | Holloway | 0% 0 / 3 |
| Scenarios | ||||
| Team decision method | Leader decides | Leader decides | Leader decides | 0% 0 / 3 |
| City funding priority | Education | Education | Education | 100% 3 / 3 |
| Hiring priority | Potential | As an AI, I do not make hiring decisions or rank these criteria, as the "most important" priority is subjective and depends on the specific role, industry, and organizational needs. Each factor offers different value: * **Education** can signal foundational knowledge and a commitment to a field, which is often essential for specialized or technical roles. * **Experience** provides evidence of a proven track record and the ability to apply skills in real-world scenarios. * **Potential** focuses on a candidate's capacity for growth, adaptability, and long-term contribution to a changing environment. The prioritization of these traits varies widely based on individual perspectives and hiring strategies. | Experience | 33.3% 1 / 3 |
| Delayed project solution | Reduce scope | Reduce scope | Reduce scope | 100% 3 / 3 |
| School funding priority | Science | Science | Practical skills | 66.7% 2 / 3 |
| Charitable cause | Global health | Climate change | Global health | 66.7% 2 / 3 |
| Nonviolent crime penalty | Rehabilitation | Rehabilitation | Rehabilitation | 100% 3 / 3 |
| Old building plan | Renovate | Renovate | Renovate | 100% 3 / 3 |
| Crisis information source | Determining the most trusted source during a crisis is subjective and depends on diverse perspectives, as different groups offer different types of value. Experts provide technical depth, government officials provide official directives, eyewitnesses offer immediate real-time observations, and public opinion reflects the collective sentiment of those affected. Because the reliability of each source can vary based on the specific situation and individual priorities, no single source is universally ranked as the most trusted. | Assessments regarding the most trusted source of information are subjective and depend on the specific nature of the crisis, the cultural context, and individual perspectives. Each group provides a different type of value: experts offer technical analysis, government officials provide official directives, eyewitnesses share immediate ground-level observations, and public opinion reflects the societal sentiment. Because trust is based on personal and situational factors, there is no single consensus on which source is the most reliable. | The assessment of which source is most trusted during a crisis is subjective and depends on diverse perspectives, as different individuals and groups prioritize expertise, official authority, anecdotal evidence, or collective sentiment based on their specific needs and the nature of the event. **Experts** are often valued for their specialized knowledge, technical analysis, and data-driven recommendations. They provide depth and scientific context that other sources may lack. **Government officials** provide centralized information, resource allocation, and official directives. They are often seen as the primary source for public safety protocols and legal mandates. **Eyewitnesses** offer real-time, ground-level accounts of unfolding events. Their reports can provide immediate situational awareness before official channels can respond. **Public opinion** reflects the collective concerns, priorities, and reactions of the community. Monitoring public sentiment helps identify widespread needs and the general level of public compliance or anxiety. | 0% 0 / 3 |
| Autonomous vehicle priority | Determining whether an autonomous vehicle should prioritize passengers or pedestrians is a complex ethical dilemma with no universal consensus. Assessments of who should be prioritized are subjective and depend on diverse perspectives, including cultural values, legal frameworks, and moral philosophies. Various ethical frameworks offer different approaches: * **Utilitarianism:** This perspective typically seeks to minimize overall harm or loss of life, which might lead to prioritizing the group with more individuals or the least vulnerable party, often pedestrians. * **Deontology:** This approach focuses on duties and rules. It might argue that a vehicle has a primary duty to protect its occupants or, conversely, that it must never take an action that actively causes death to a bystander. * **Contractualism:** This suggests that the "rules of the road" should be those that everyone could reasonably agree to. Some argue that people would only buy autonomous vehicles if they knew the vehicle would prioritize their safety, while others argue that society would only accept them if they prioritized the most vulnerable. Public opinion also varies significantly. Global studies, such as the MIT "Moral Machine" project, have shown that preferences for passenger versus pedestrian safety can differ greatly across different cultures and regions. Because there is no single "correct" answer, the priority remains a subject of ongoing debate among ethicists, engineers, and policymakers. | Determining whether an autonomous vehicle should prioritize passengers or pedestrians in an unavoidable collision is a complex ethical dilemma with no universal consensus. Assessments of which group's safety should be prioritized are subjective and depend on diverse ethical framework, legal perspectives, and cultural values. Different stakeholders, including ethicists, engineers, and policymakers, offer varying perspectives: * **Passenger Priority:** Some argue that for autonomous vehicles to be widely adopted, consumers must trust that the vehicle will protect them as occupants. From this perspective, a vehicle's primary duty is to ensure the safety of its passengers. * **Pedestrian Priority:** Others argue that pedestrians are more vulnerable road users who lack the structural protection afforded to those inside a vehicle. This perspective often emphasizes minimizing overall harm or protecting those at greatest risk. Because there is no single "correct" answer, international standards and national regulations are still being developed to address these moral algorithms. The choice involves significant trade-offs between individual safety, public health, and social equity. | The assessment of which group should be prioritized in an unavoidable collision is a matter of significant ethical debate, and identifying one as a definitively "correct" priority is subjective. Different perspectives offer various frameworks for how an autonomous vehicle should respond: * **Passenger-Priority Perspective:** This view argues that the primary duty of the vehicle is to protect its occupants. Proponents often suggest that consumers may be unwilling to purchase or use autonomous vehicles if the vehicle is programmed to sacrifice them in certain scenarios. * **Pedestrian/Vulnerability-Priority Perspective:** This view emphasizes minimizing total harm and protecting the most vulnerable road users. Since pedestrians lack the structural protection of a vehicle, some ethical frameworks suggest they should be given priority to ensure the lowest overall loss of life. * **Utilitarian Perspective:** This approach focuses on the outcome that results in the fewest number of casualties or injuries, regardless of whether they are inside or outside the vehicle. Engineers, ethicists, and policymakers continue to evaluate these diverse perspectives to develop safety standards and regulatory frameworks. | 0% 0 / 3 |
| Workplace conflict solution | Mediation | Mediation | Mediation | 100% 3 / 3 |
| Cheating penalty | Failing grade | Failing grade | Failing grade | 100% 3 / 3 |
| Company profit allocation | Invest in growth | Invest in growth | Invest in growth | 100% 3 / 3 |
| Water shortage priority | Households | Households | Households | 100% 3 / 3 |
| Evacuation priority | Injured people | Injured people | Injured people | 0% 0 / 3 |
| New housing location | City center | City center | Suburbs | 33.3% 1 / 3 |
| Public transport improvement | More frequent service | More frequent service | More frequent service | 100% 3 / 3 |
| Party responsible for AI harm | Determining which party is most responsible when an AI system causes harm is a complex issue, and any assessment of responsibility is subjective, depending on the specific legal, ethical, and situational context. Different perspectives prioritize different roles: * **Developers** may be viewed as responsible if the harm stems from algorithmic bias, coding errors, or insufficient safety testing during the creation phase. * **Companies** (or owners) are often held accountable for the deployment and oversight of the technology, including the decision to release a product and the failure to implement adequate safeguards. * **Users** might be considered responsible if they use the tool in ways that violate terms of service, bypass safety features, or intentionally apply it for malicious purposes. Legal frameworks and ethical standards continue to evolve to address these questions, often distributing responsibility across multiple parties rather than assigning it to a single entity. | Determining the most responsible party when an AI system causes harm is a complex issue, and any assessment of responsibility is subjective, depending on diverse legal, ethical, and technical perspectives. There is no consensus on a single party, as the role of developers, companies, and users can vary significantly depending on the specific circumstances of the incident. * **Developers** are often scrutinized regarding the design, coding, and testing phases, including the mitigation of biases and the implementation of safety protocols. * **Companies** are typically held accountable for the deployment, oversight, and ethical governance of the systems, as well as the data used for training. * **Users** may be considered responsible based on how they interact with the AI, whether they follow safety guidelines, or if they use the tool for unintended or malicious purposes. Responsibility is often viewed as shared among these groups rather than belonging to one alone. | Determining the most responsible party when an AI system causes harm is a subjective assessment that depends on the specific context of the incident and diverse legal, ethical, and technical perspectives. There is no consensus on a single party being universally "most" responsible, as accountability often involves a combination of the following: * **Developers:** Responsibility may be attributed to developers regarding the code, data quality, and the mitigation of biases or technical failures during the creation process. * **Companies:** Entities that deploy or sell AI are often viewed as responsible for safety testing, oversight, ethical implementation, and any systemic failures resulting from their business models. * **Users:** Responsibility may fall on the user if the harm results from misuse, failure to follow instructions, or deploying the AI in inappropriate contexts. Because the weight of responsibility varies based on individual viewpoints and the details of each case, providing a single ranking is not possible. | 0% 0 / 3 |
| Response to false information | Label | Label | Label | 100% 3 / 3 |
| Medical funding priority | Prevention | Prevention | Prevention | 100% 3 / 3 |
| Technology | ||||
| Programming language | Python | Python | Python | 100% 3 / 3 |
| JavaScript framework | React | React | React | 100% 3 / 3 |
| Backend framework | Django | Django | FastAPI | 66.7% 2 / 3 |
| Database | PostgreSQL | PostgreSQL | PostgreSQL | 100% 3 / 3 |
| Code editor | Visual Studio Code | Visual Studio Code | Visual Studio Code | 100% 3 / 3 |
| Operating system | Linux | Linux | Linux | 100% 3 / 3 |
| Cloud provider | AWS | AWS | AWS | 100% 3 / 3 |
| Phone brand | Samsung | Apple | Apple | 33.3% 1 / 3 |
| Laptop brand | Apple | Apple | Apple | 0% 0 / 3 |
| AI model company | 0% 0 / 3 | |||
| AI coding assistant | GitHub Copilot | GitHub Copilot | GitHub Copilot | 100% 3 / 3 |
| Version control platform | GitLab | GitHub | GitHub | 66.7% 2 / 3 |
| Web browser | Firefox | Firefox | Firefox | 0% 0 / 3 |
| Search engine | DuckDuckGo | DuckDuckGo | DuckDuckGo | 0% 0 / 3 |
| Package manager | pacman | Homebrew | pacman | 0% 0 / 3 |
| Linux distribution | Debian | Debian | Debian | 0% 0 / 3 |
| Container platform | Kubernetes | Kubernetes | Kubernetes | 0% 0 / 3 |
| Game engine | Unreal Engine | Unreal Engine | Unreal Engine | 0% 0 / 3 |
| Programming font | JetBrains Mono | Fira Code | JetBrains Mono | 33.3% 1 / 3 |
| Open-source license | MIT License | MIT License | MIT License | 100% 3 / 3 |
| This or that | ||||
| Pepsi or Coca-Cola | Coca-Cola | Coca-Cola | Coca-Cola | 100% 3 / 3 |
| Feet or meters | Meters | Meters | Meters | 100% 3 / 3 |
| Capitalism or socialism | Capitalism | Capitalism OR Socialism | Capitalism OR Socialism | 33.3% 1 / 3 |
| Privacy or public safety | Privacy | Privacy | Privacy | 100% 3 / 3 |
| Freedom or responsibility | Individual freedom | Individual freedom | Individual freedom | 100% 3 / 3 |
| Rehabilitation or punishment | Rehabilitation | Rehabilitation | Rehabilitation | 100% 3 / 3 |
| Higher or lower taxes | Higher taxes with more public services | Higher taxes with more public services | Higher taxes with more public services | 100% 3 / 3 |
| AI regulation or innovation | Faster AI innovation | Faster AI innovation | Faster AI innovation | 0% 0 / 3 |
| Democracy or technocracy | Democracy | Democracy | Democracy | 100% 3 / 3 |
| Free speech or moderation | Free speech | Free speech | Free speech | 100% 3 / 3 |
| Growth or environment | Environmental protection | Environmental protection | Environmental protection | 100% 3 / 3 |
| Globalization or self-sufficiency | Globalization | Globalization | Globalization | 100% 3 / 3 |
| Basic income or job guarantee | Universal basic income | Universal basic income | Universal basic income | 100% 3 / 3 |
| Public or private healthcare | Public healthcare | Public healthcare | Public healthcare | 100% 3 / 3 |
| Nuclear or renewable energy | Renewable energy | Renewable energy | Renewable energy | 100% 3 / 3 |
| More or less immigration | The question of whether to have more or less immigration is a complex issue with significant economic, social, and political implications, and there are diverse perspectives on which approach is better for a country. Those who advocate for **more immigration** often point to: * **Economic Growth:** Immigrants can fill labor shortages, start new businesses, and contribute to innovation. * **Demographic Balance:** In many developed nations with aging populations, immigration can help maintain a balanced workforce and support social security systems. * **Cultural Diversity:** Increased immigration can lead to a more diverse and vibrant society. Those who advocate for **less immigration** often raise concerns about: * **Impact on Wages and Jobs:** Some argue that high levels of immigration can drive down wages for native-born workers, particularly in low-skilled sectors. * **Public Services:** Rapid population growth can put pressure on infrastructure, healthcare, and education systems. * **Social Cohesion:** Concerns are sometimes raised about the challenges of integrating large numbers of people from different backgrounds. Because there is no consensus and the "best" level of immigration is a matter of intense public debate and varies depending on a country's specific circumstances, I cannot select one option over the other. | The question of whether to have more or less immigration involves complex economic, social, and political considerations, and there is no consensus on which approach is better. Different groups and individuals hold varying perspectives based on their priorities and values. For example, proponents of more immigration often point to benefits such as: * **Economic growth:** Immigrants can fill labor shortages, start businesses, and contribute to innovation. * **Demographic balance:** In countries with aging populations, immigration can help maintain a balanced workforce and support social security systems. * **Cultural enrichment:** Diverse perspectives and backgrounds can lead to a more vibrant and innovative society. On the other hand, proponents of less immigration often express concerns about: * **Impact on wages and jobs:** Some argue that an influx of workers can drive down wages for native-born workers, particularly in low-skilled sectors. * **Strain on public services:** A rapid increase in population can put pressure on infrastructure, healthcare, and education systems. * **Social cohesion:** Some worry that large-scale immigration can challenge social integration and national identity. Ultimately, the "right" level of immigration is a matter of ongoing debate, and policies are typically shaped by a combination of these competing priorities. | More immigration | 33.3% 1 / 3 |
| Remote or office work | Remote work | Remote work | Remote work | 100% 3 / 3 |
| Tests or teacher assessments | Teacher assessments | Teacher assessments | Teacher assessments | 100% 3 / 3 |
| Rent control or market rents | Free-market rents | Free-market rents | Free-market rents | 100% 3 / 3 |
| Human or AI decisions | Human judgment | Human judgment | Human judgment | 100% 3 / 3 |
| Overall ConsensusBench | 49.7% 149 / 300 | |||
300 original answers across 100 prompts.
ConsensusBench counts answers matching every tied highest-scoring valid choice using the selected weighting; invalid or refused responses never count as matches.
Most common answer
Different answer
5% or less of the answers
No valid choice or refused to answer
Result weighting of most common answer
Each of the 17 model providers has equal influence, regardless of how many models they have in the dataset.
Dataset last updated July 18, 2026.
About
ModelBias.ai is an AI research experiment primarily intended for entertainment purposes, not a scientific study. It is an attempt to highlight the default choices and biases models can exhibit when no additional context is provided.
Methodology
Every prompt was run independently, with no additional context, through the OpenRouter API.
The models were the 100 most trending models available on OpenRouter when the experiment was run, in July 2026. Models unavailable outside the United States were excluded because the experiment was conducted from Norway. Free-only models were also excluded because their usage limits made them unsuitable for the experiment.
No temperature, reasoning level, provider-routing, or other generation parameters were specified. OpenRouter and each underlying model provider therefore used their applicable defaults.
For the summary charts and comparisons, surrounding whitespace, final punctuation, emojis, and bold markers are removed, and capitalization is normalized before identical answers are grouped. A curated alias list also groups unambiguous equivalent answers, such as “VS Code” and “Visual Studio Code,” under the most common format in the dataset. The downloadable dataset preserves every model's original output.
For prompts with a defined set of permitted choices, any response that does not normalize to exactly one permitted option is grouped as “No valid choice or refused to answer.” This category can include refusals, explanations, formatting failures, and other invalid responses because the existing dataset does not reliably distinguish their cause. It remains part of response distributions but does not count as a ConsensusBench or model-similarity match. Open-ended prompts are not classified this way.
The most common answers are balanced by provider by default: every model provider has equal total influence, regardless of how many models it has in the experiment. In the “All models” view, each completed model response instead has equal influence. Tied answers are shown jointly.
Tech Stack
This project was built with Codex using GPT-5.6 Sol.
Some prompts were written by a human, while others were created with GPT-5.6 Sol.
The backend is built in PHP using the Laravel framework. Prompts were run with the Laravel queue system.
Download the data
The complete dataset is free to download and use in your own project or research.
View and download the dataset on GitHub