Benchmark Almanac

Emergent Price-Fixing by LLM Auction Agents

Grok 3 19% leads since July 2025
Not in model cards

Models trade in simulated auctions and communicate with other participants. A judge examines game logs for evidence of price fixing.

The author reports the share of games with an illegality score of at least seven for 13 model configurations.

Made by
Lech Mazur
Released
July 2025
Kind
Deception and alignment
Level
Everyday
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Measures
Propensity
Behavior
Collusion
Language
English
In and out
Text → Actions, Text
Interaction
Multi-agent
Environment
A simulator
Items from
Generated
Refresh
One fixed set
Run by
Anyone
In views
Agents, Robotics
Strongest model tested
o3 #84 on the capability index
Last model added
July 2025
Scores refreshed
November 21st, 2026
13
Entries in its paper, 12 models we track

Grok 3 leads with 19% among models released by July 2025 Grok 3 leads with 19% among models released by July 2025

Best score per entry from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

The newest model with a score here came out in July 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 Grok 3 no reasoning 19%
  2. 2 GPT-4o Mar 2025 23%
  3. 3 DeepSeek-V3 Mar 2025 26%
  4. 4 Grok-3 mini high-reasoning 30%
  5. 5 Claude Sonnet 4 no reasoning 32%
  6. 6 Claude Opus 4 no reasoning 36%
  7. 7 Gemini 2.5 Pro Jun 2025 39%
  8. 8 Mistral Medium 3 47%
  9. 9 o3 51%
  10. 10 Qwen-3 235B 57%
  11. 11 o4-mini 62%
  12. 12 DeepSeek-R1 May 2025 71%
  13. 13 Grok 4 75%

Source: The Emergent Price-Fixing by LLM Auction Agents paper

#EntryScoreSettingDateSource
1 Grok 3 19% no reasoning Apr 2025 Official leaderboard
2 GPT-4o Mar 2025 23% Mar 2025 Official leaderboard
3 DeepSeek-V3 Mar 2025 26% Mar 2025 Official leaderboard
4 Grok-3 mini 30% high-reasoning Apr 2025 Official leaderboard
5 Claude Sonnet 4 32% no reasoning May 2025 Official leaderboard
6 Claude Opus 4 36% no reasoning May 2025 Official leaderboard
7 Gemini 2.5 Pro Jun 2025 39% Jun 2025 Official leaderboard
8 Mistral Medium 3 47% May 2025 Official leaderboard
9 o3 51% Apr 2025 Official leaderboard
10 Qwen-3 235B 57% Official leaderboard
11 o4-mini 62% Apr 2025 Official leaderboard
12 DeepSeek-R1 May 2025 71% May 2025 Official leaderboard
13 Grok 4 75% Jul 2025 Official leaderboard
#ModelScoreSettingDateSource
1 Grok 3 19% no reasoning Apr 2025 Official leaderboard
2 GPT-4o Mar 2025 23% Mar 2025 Official leaderboard
3 DeepSeek-V3 Mar 2025 26% Mar 2025 Official leaderboard
4 Grok-3 mini 30% high-reasoning Apr 2025 Official leaderboard
5 Claude Sonnet 4 32% no reasoning May 2025 Official leaderboard
6 Claude Opus 4 36% no reasoning May 2025 Official leaderboard
7 Gemini 2.5 Pro Jun 2025 39% Jun 2025 Official leaderboard
8 Mistral Medium 3 47% May 2025 Official leaderboard
9 o3 51% Apr 2025 Official leaderboard
10 o4-mini 62% Apr 2025 Official leaderboard
11 DeepSeek-R1 May 2025 71% May 2025 Official leaderboard
12 Grok 4 75% Jul 2025 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

xAI
Benchmark paper Not tested since Jul 2025
Grok 3 19%

Source: The Emergent Price-Fixing by LLM Auction Agents paper

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
In the benchmark’s paper Best so far
0% 20% 40% 60% 80% 100% 2025 Grok 3 DeepSeek-V3 0% 20% 40% 60% 80% 100% 2025 Grok 3 DeepSeek-V3

Source: Official Emergent Price-Fixing by LLM Auction Agents leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Safety and alignment

All 636 in Safety and alignment
  • FORTRESS In use Paired adversarial and benign national-security misuse prompts. Rubric-graded risk score on adversarial and over-refusal score on benign prompts. Claude Opus 5.5 5.78%
  • StrongREJECT In use Forbidden-prompt jailbreak benchmark with a rubric grader that scores how useful a non-refusal actually is. Inkling 98.6%
  • Gray Swan IPI Arena benchmark In use Indirect prompt-injection attacks from public red-teaming competitions replayed on tool, coding and browser agents. Attack success rate at k attempts. Gemini 4 Carbon 0%
  • Covert misbehavior (Thinking Machines) In use Agentic tasks that check whether a model misbehaves and conceals it. Scored by the rate of covert misbehavior. Inkling 2 Mini 8.30%
  • Destructive actions (Thinking Machines) In use Agentic tasks that check whether a model takes destructive actions it was not authorized to take. Scored by the rate of destructive actions. Inkling 2 Small 4.70%
  • Misbehavior (Thinking Machines) In use Agentic tasks that check whether a model fabricates results, ignores instructions or hides what it did when a task is hard. Scored by the rate of misbehavior. Inkling 2 Mini 9.80%

Data audit

8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 16 months ago, in July 2025, for Grok 4. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
  • Pass Independent runs Scores from the benchmark’s paper (13 entries).
  • Pass Models scored 12 models we track, plus 1 other entry on its leaderboard.
  • Warning Newest score The newest came 16 months ago, in July 2025, for Grok 4.
  • Note Frontier models None of the top 10 has a score. The strongest tested is o3 (#84).
  • Note Headroom Lower is better, so there’s no ceiling to reach.
Leaderboard import 3 pass
  • Pass Importer We read the board on github.com: 12 rows for models we track and 1 other entry at the last read.
  • Pass Last read November 21st, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check