Skip to content
Apart Research
Apart Research
Contact us
Sprint projectOct 26, 2026Online(from Kerala, India)

An LLM Pricing Hub Survives Data Limits on Some Models; Firm Discretion Works Only When Firms Refuse Small Nudges

Ebin Babu Thomas

Submitted to AI Collusion Research Sprint. Sprint projects are early-stage work by participants, not Apart Research publications.

Read the report

Report: An LLM Pricing Hub Survives Data Limits on Some Models; Firm Discretion Works Only When Firms Refuse Small Nudges

Open in a new tab (opens in new tab)
Presentation

Presentation: An LLM Pricing Hub Survives Data Limits on Some Models; Firm Discretion Works Only When Firms Refuse Small Nudges

Open in a new tab (opens in new tab)
Code (opens in new tab)More on github.com (opens in new tab)
Share

We test whether the remedies in the DOJ's 2026 RealPage Final Judgment stop an LLM pricing hub that advises competing firms and is instructed to maximise their combined profit, the incentive alleged against RealPage. In a repeated Bertrand duopoly (700+ runs, seven models from six labs, main batch pre-registered), the hub drives a competitive market to joint monopoly. With a per-client objective and the same rival data it is pro-competitive: the objective, not the data, does the harm. A data ban removes most of the harm on three models and almost none on two. Per-round auto-accept bands only delay monopoly. Firm discretion works only when the deciding firm refuses small nudges: a scripted one-cent-per-round ratchet takes three of five firm models most or all of the way to monopoly, and a monitor of single price moves misses these climbs. Remedies should target the advisor's objective and audit acceptance rates and where prices settle.

Reviews

Judging this Sprint?

Review “An LLM Pricing Hub Survives Data Limits on Some Models; Firm Discretion Works Only When Firms Refuse Small Nudges”

Your public critique appears on the project page without your name. Reviews of strong projects are shared with funders, as you choose below.

Not shown on the project page.

Shown on the project page, without your name.

Only the Apart team reads this, and funders if you agree below.

Decline to score

Only when the quality of the writing makes the work impossible to follow. Say why in your private critique; the project page shows only that a judge declined, not why.

How much would this matter for the field if it worked? How innovative is it? For scores of 4-5: is this actually new to the field, or replicating recent work?

Scoring guide
  1. 1Negligible. No clear problem addressed, or no meaningful novelty.
  2. 2Limited. Addresses a real problem but with a generic or well-trodden approach. Incremental at best.
  3. 3Moderate. Clear problem with a reasonable approach; some novelty in framing or method beyond routine application of existing tools.
  4. 4Significant. Important problem with an original approach, or identifies a neglected problem area. A valuable contribution others could build on.
  5. 5Exceptional. Tackles a critical problem with a genuinely novel approach, or opens a new research direction. Clear theory of change. You'd be excited to share this with researchers in the area.

How sound are methodology, implementation, and findings?

Scoring guide
  1. 1Seriously flawed. Methodology broken, results uninterpretable, or implementation doesn't work.
  2. 2Weak. Approach has significant gaps: missing validation, flawed experimental design, or incomplete implementation.
  3. 3Competent. Technically solid given the short duration. Methodology makes sense, results are interpretable, limitations acknowledged, work builds toward clear conclusions.
  4. 4Strong. Thorough methodology with convincing validation. Results clearly support conclusions. Immediately useful for future work.
  5. 5Exceptional. Ambitious scope executed rigorously. Surprising findings, novel methods, or unusually robust validation.

How clearly are work, findings, and impact potential communicated?

Scoring guide
  1. 1Incomprehensible. Cannot determine what the project is actually claiming or doing.
  2. 2Hard to follow. Key information buried, missing, or diluted by excessive length. Significant effort to extract main points.
  3. 3Clear enough. Can understand the problem, approach, and results without undue effort. Core content clearly present: problem, method, findings, limitations.
  4. 4Well presented. Easy to follow, well-structured, appropriate level of detail. Target audience would get it quickly.
  5. 5Exceptionally clear. A pleasure to read. Complex ideas made accessible. Could serve as a model for how to present this type of work.
Sharing your name *

When we share a strong project with funders, your name, scores and public critique go with it, confidentially; this choice is only about whether your name is public. If you think this project deserves funding, please also endorse it: that carries the most weight.

Share my private critique with funders *

See our Privacy Policy.

  1. This experiment sheds light on the efficacy of the proposed remedies in the US DOJ's settlement of the RealPage case, which some have criticized as too lenient. In the context of the posited LLM pricing hub, the experiment provides support for that criticism. Of particular note is the finding that it is the objective, not the data, that drives the duopoly in the experiment towards monopoly pricing in a Bertrand model.

  2. This project tests the remedies based on the judgment of the DOJ–RealPage case on collusion with a price recommender LLM. The premises of the study, such as what is considered collusion, are clearly defined. The project uses the logit Bertrand duopoly model as the economic model between sellers, which is standard and consistent with the existing literature. The clarity of the report still has room to improve. The results of the study have some practical implications and can give the public some understanding of the effectiveness of the remedies.

  3. Taking the actual remedies from the DOJ's RealPage settlement and testing them directly in an LLM hub-and-spoke setup is a great paper idea. The takeaway that the hub's objective function matters way more than data-sharing limits is something antitrust folks actually need to hear right now.

    The main weak spot is statistical power in a few of your replication arms, with the low samples sizes (n=3 to 12) bring pretty pretty thin to lean on heavily. I'd also be curious how much of the "joint-profit" behavior survives if you don't explicitly hand the hub a joint-profit prompt, or if humans are the ones taking the small nudges instead of LLM spoke agents.

Cite this project

@misc{thomas2026llm,
  title = {{An LLM Pricing Hub Survives Data Limits on Some Models; Firm Discretion Works Only When Firms Refuse Small Nudges}},
  author = {Ebin Babu Thomas},
  year = {2026},
  month = oct,
  note = {Submitted to AI Collusion Research Sprint, an Apart Research Sprint},
  howpublished = {\url{https://apartresearch.com/sprints/projects/an-llm-pricing-hub-survives-data-limits-on-some-models-firm-discretion-works-only-when-firms-refuse-small-nudges-6cca}},
  url = {https://apartresearch.com/sprints/projects/an-llm-pricing-hub-survives-data-limits-on-some-models-firm-discretion-works-only-when-firms-refuse-small-nudges-6cca}
}

More from AI Collusion Research Sprint

All 392 entriesBrowse all projects

A Sprint is running now

AI x Epistemics Research Sprint · Nov 13 - 15, 2026