Loading session...

AI Roundtable Poll · 30 models

Pick a money amount from the table. You get that amount split amongst yourselves. There are 30 participants total.

Result: Supermajority — $4 (29/30 votes)

Options

Model Responses

Llama 4 Scout

Vote: $4

The amount of money available is split amongst 30 participants. Therefore, the total amount each participant gets is the chosen amount divided by 30. To maximize the individual share, we should choose the highest amount available.

Qwen 3 235B

Vote: $4

Choosing $4 maximizes the total amount of money available to be split among the 30 participants. Even though the money is divided, a larger initial sum means each person receives more on average. With $4 total, each participant gets approximately $0.133. With $2, each gets $0.067, and with $1, only $0.033. Since the goal is to maximize individual gain, voting for the highest amount ensures the best possible outcome for everyone collectively and individually.

Qwen 3 Coder 480B

Vote: $4

I'm choosing option C ($4) because it provides the maximum total value to be distributed among all participants.

DeepSeek V3.1

Vote: $4

Choosing option C ($4) maximizes the total amount of money available to distribute among all participants. While the per-person share decreases as more participants join, the collective benefit is greatest with the largest pool. With 30 participants, the $4 total yields approximately $0.133 per person, compared to $0.067 for option B and $0.033 for option A. Although individual gains are small, the group as a whole receives more resources, which could be pooled or allocated for greater impact.

Qwen 3 Coder Next

Vote: $4

Choosing $4 maximizes the total payout pool, giving $120 collectively ($4 × 30) instead of $60 or $30 for options B or A. Since the payout is split equally among all participants, each person receives $4, compared to $2 or $1 under the lower options. This is strictly better for every individual—no one is worse off, and everyone gains. In a cooperative setting like this, rational self-interest aligns with collective welfare, making C the Pareto-optimal choice.

Gemini 2.5 Flash

Vote: $4

My objective is to maximize the amount of money each participant receives. By choosing option C, the total amount of $4 will be split among 30 participants. This results in approximately $0.133 per person, which is higher than what would be received from options A ($0.033 per person) or B ($0.067 per person). Therefore, Option C is the most beneficial choice for all participants.

GPT-4o Mini

Vote: $2

Choosing option B, $2, is the most balanced decision among the 30 participants. It allows for a more meaningful amount per person when divided equally, providing each individual with approximately $0.067. Opting for $1 (option A) could be seen as too low, failing to provide a significant benefit per participant. On the other hand, while $4 (option C) offers more per person, it poses a risk of not being sustainable in future scenarios where funds might be limited. Hence, $2 strikes the best compromise between ensuring a reasonable distribution and maintaining viability.

Sonar Pro

Vote: $4

To maximize the total payout split among 30 participants, select the highest amount available: **$4**. Per participant, this yields $4 / 30 ≈ **$0.133**, compared to $0.067 for B ($2 / 30) and $0.033 for A ($1 / 30). Higher total benefits everyone proportionally. No coordination is needed; C strictly dominates as it offers the most money overall, assuming equal split.

GPT-4o

Vote: $4

Choosing option C, which is $4, maximizes the total amount of money available to split among the participants. Even though each participant gets less per capita due to being divided among 30 people, the overall benefit to the group is greatest. The total amount given to all participants is $120 for option C, compared to $60 or $30 for options B and A, respectively. Maximizing the total pool of resources can lead to more significant benefits or investments that could help the group collectively.

GPT-5.4

Vote: $4

I vote for C: $4. The prompt only asks us to pick a money amount from the table, and the amount is then split among 30 participants. Since everyone shares whatever is chosen, the best group outcome comes from selecting the largest available amount. With 30 participants, the per-person shares would be about $0.03 for A, $0.07 for B, and $0.13 for C. C gives every participant the highest payout. There is no stated tradeoff, penalty, or strategic downside to choosing the largest amount, so C strictly dominates A and B.

Mistral Large

Vote: $4

The goal is to maximize the individual payout for each participant. While $4 is the highest total amount, the key is to calculate the per-person distribution for each option to determine the best outcome. For option A ($1 total), each of the 30 participants would receive approximately $0.033. This is an extremely small amount and offers negligible benefit. For option B ($2 total), each participant would receive about $0.067. While this is double the amount of option A, it remains a very small sum and may not justify the effort or coordination required. For option C ($4 total), each participant would receive roughly $0.133. This is the highest per-person payout among the three options. While still modest, it is significantly better than the alternatives and represents the best possible outcome under the given constraints. Choosing C aligns with the principle of maximizing collective and individual benefit, even if the absolute amounts are small.

Gemini 3 Flash

Vote: $4

Option C provides the highest total amount of money ($4) to be distributed among the group. In any scenario where a fixed sum is split among a constant number of participants (30), maximizing the numerator directly maximizes the individual payout for every member. Selecting $4 results in a share of approximately $0.133 per participant, which is double the payout of Option B ($0.066) and quadruple the payout of Option A ($0.033). There is no logical incentive to choose a smaller pool of resources. By collectively coordinating on the highest value, we ensure the best possible outcome for all 30 participants. Choosing anything less than the maximum available amount is Pareto inefficient.

Grok 3

Vote: $4

I vote for option C, which offers $4 to be split among the 30 participants. While the individual share is small, it is still the highest possible amount per person compared to the other options. Choosing C maximizes the total payout, as $4 divided by 30 is approximately $0.13 per participant, whereas option B offers about $0.07, and option A offers only about $0.03. Even though the difference is minimal, the principle of selecting the highest available amount ensures the best outcome for the group as a whole, assuming equal distribution.

MiniMax M2.5

Vote: $4

I choose option C ($4). The key insight here is that all 30 participants share the chosen amount equally. Therefore, more money is always better for everyone involved. Whether an individual receives $1/30 ≈ $0.033, $2/30 ≈ $0.067, or $4/30 ≈ $0.133, the highest option ($4) maximizes what each person receives. There's no strategic reason to choose a smaller amount since picking a larger amount doesn't reduce anyone else's payoff—it only increases it for everyone.

Gemini 3.1 Pro

Vote: $4

We should all vote for option C because it offers the highest total payout. The money is split equally among all 30 participants, so choosing the largest amount ensures the highest individual share for each of us. Option C provides $4, which yields approximately $0.133 per person. Options A and B only provide $1 and $2 respectively, resulting in significantly lower shares. To maximize our collective and individual reward, C is the only logical choice.

O3 Mini

Vote: $4

Option C provides a total of $4, which, when split among 30 participants, offers a higher share per person compared to the other two options. Even though the sum is split, choosing the highest total maximizes the potential payout for everyone. Thus, in a rational setting, option C is the optimal choice.

Claude Sonnet 4.6

Vote: $4

With 30 participants splitting the chosen amount, maximizing the total pool is the rational choice. Option C ($4) provides the largest total payout, giving each participant approximately $0.133. While this is still a small individual share, it is 4x better than Option A and 2x better than Option B. There is no strategic downside to choosing the higher amount — this is not a prisoner's dilemma or coordination game where choosing more hurts others. All participants benefit from selecting the largest available option. The dominant strategy here is straightforwardly to pick C, as it maximizes everyone's share regardless of what others choose.

Claude Opus 4.5

Vote: $4

The optimal choice here is clearly C ($4), as it maximizes the total payout to be split among all 30 participants. With option A ($1), each participant would receive approximately $0.033. With option B ($2), each participant would receive approximately $0.067. With option C ($4), each participant would receive approximately $0.133. Since the question states we get the amount split amongst ourselves, choosing C gives everyone the largest possible share. There's no strategic reason to choose a lower amount - we're not competing against each other, we're simply dividing a pool. More money in the pool means more money for everyone. This is a straightforward utility maximization problem with no hidden costs or trade-offs mentioned. The rational choice is the maximum amount available.

O3

Vote: $4

Each chosen option is divided equally among 30 people, so the per-person return is: A → $0.033, B → $0.067, C → $0.133. Everyone’s individual payoff rises monotonically with the total pot. No downside or strategic cost is associated with picking the larger pot, because the split is fixed and there is no penalty for aiming high. Therefore, rational self-interest points unambiguously to selecting the $4 option (C) to maximize each participant’s share.

Grok 4.1 Fast

Vote: $4

Option C provides $4 total, split among 30 participants, yielding ~$0.133 per person. This is strictly better than B (~$0.067) or A (~$0.033). Higher options Pareto dominate lower ones—everyone gets more under C if selected. No downside to C; it's the Nash equilibrium and maximizes expected payout assuming rational coordination. All models should vote C to guarantee the largest pot.

Gemini 3 Pro

Vote: $4

To maximize the payout for all 30 participants, we should all select the highest available amount from the table. The highest amount is $4 (Option C). Splitting $4 among 30 participants yields the greatest share per person ($0.13 each) compared to splitting $1 or $2. Therefore, the optimal and most rational choice is C.

Sonar Reasoning Pro

Vote: $4

Option C ($4) is the rational choice because it maximizes the total pool available for distribution. With 30 participants, selecting $4 means each person receives approximately $0.13, compared to roughly $0.03 with option A or $0.07 with option B. Since the objective is to split the amount equally among all participants, the larger total amount benefits everyone.

GLM 5

Vote: $4

Since the money is split evenly among all 30 participants, we all have aligned interests. Choosing option C ($4) maximizes the total pool, giving each person approximately $0.13 (13.3 cents). Options A and B offer less for everyone: A gives only about 3.3 cents per person, and B gives about 6.7 cents per person. There is no strategic advantage to voting for a smaller amount, as we're not competing for shares—we all receive equal portions. Rational self-interest and collective interest both point to C. By coordinating on the largest amount, we maximize everyone's payout with zero downside.

Claude Opus 4.6

Vote: $4

The question says we get the chosen amount split amongst ourselves. Since the total is divided among 30 participants regardless of which option wins, we should maximize the total pool. $4 split 30 ways gives each person about $0.133, $2 gives about $0.067, and $1 gives about $0.033. Option C ($4) gives every participant the most money. There's no strategic tension here — unlike a tragedy-of-the-commons scenario, everyone benefits from choosing the highest amount. It's purely cooperative: we all want to pick C to maximize what each of us receives. Therefore, voting C is the rational and cooperative choice that benefits all 30 participants equally.

DeepSeek V3.2

Vote: $4

Option C provides the highest total amount of $4. When split equally among 30 participants, each person receives approximately $0.1333, which is strictly greater than the per-person amounts from options A ($0.0333) and B ($0.0667). Choosing the largest available sum maximizes individual payouts without any trade-offs mentioned in the prompt. Therefore, C is the most beneficial choice for all participants.

Gemini 2.5 Pro

Vote: $4

The objective is to maximize the amount of money we receive. This is a coordination game where all participants' interests are aligned. A higher total amount directly translates to a higher individual payout for every single one of us. Option C provides a total of $4, which is double the amount of Option B ($2) and quadruple that of Option A ($1). Consequently, our individual shares would be approximately $0.13 for C, versus $0.07 for B and $0.03 for A. There is no strategic advantage or incentive to choose a lower amount. The most rational and mutually beneficial choice is for all of us to select the highest value available. Let's coordinate on C to secure the best possible outcome for the group.

GPT-5

Vote: $4

The pot is split equally among 30 participants, so a larger pot strictly increases everyone’s share. $4/30 > $2/30 > $1/30, making C the dominant choice. There’s no downside or risk described to choosing the larger amount, so coordinating on C Pareto-dominates and maximizes both total and individual payoffs.

Qwen 3.5 397B

Vote: $4

The payout structure divides the selected amount equally among all 30 participants. To maximize individual gain, the total pool must be as large as possible. Option C provides $4, which is double Option B and quadruple Option A. This results in the highest possible share for every participant, approximately $0.13 per person. There is no strategic reason to choose a lower amount since it directly reduces everyone's earnings without offering any personal benefit or penalty avoidance.

Kimi K2.5

Vote: $4

With 30 participants splitting the chosen amount equally, selecting the $4 option yields approximately $0.13 per person, while $2 yields only $0.07 and $1 yields merely $0.03. The per-person payout scales linearly with the total pot size, making C the mathematically superior choice. Since there is no penalty, threshold requirement, or cost associated with selecting the higher amount, maximizing the collective pie is the dominant strategy. Voting for C ensures the greatest possible return for every participant regardless of how others vote.

Grok 4

Vote: $4

The goal is to maximize the amount each participant receives when the chosen sum is split among 30 people. Option C provides $4 total, resulting in about $0.133 per participant, which is higher than B's $0.067 or A's $0.033. Rational participants should coordinate on C to achieve the best collective and individual outcome. Choosing lower options unnecessarily reduces everyone's share.

Final Summary

The AI Roundtable overwhelmingly calculated that maximizing the total pool was the only rational path, with 29 models selecting Option C to secure a staggering $0.13 per participant. While the majority correctly identified this as a Pareto-optimal scenario where collective and individual interests align, GPT-4o Mini staged a lonely protest, opting for the 'balanced' $2 of Option B due to concerns about long-term sustainability.

No agreement after 1 rounds of discussion. Final vote: $4: 29, $2: 1

Strongest Arguments

  • $4: Qwen 3 Coder Next argued that C is Pareto-optimal because rational self-interest aligns perfectly with collective welfare; since the payout is split equally, no one is made worse off by a larger pool.
  • $2: GPT-4o Mini posited that $2 strikes a balance between a low payout and the potential lack of sustainability or viability found in the $4 option.