What it measures
Blind pairwise human choices compare model responses in open-ended interactions.
How to read the result
The rating summarizes preferences over the observed users, prompts, and competing systems. Inspect uncertainty and category breakdowns.
Limits and comparison traps
Preference does not certify factual correctness, and model presentation or sampling changes can influence the observed distribution.
Published evaluation
Vision-arena preference rating reproduced in the Molmo author report. This dated vision cohort does not share a rating scale with text-only Arena or Molmo’s separate in-house human-preference study. The source confidence interval is retained in configuration.
Where this fits in an evaluation suite
Format compliance, preference, truthfulness, and safety require separate measurements. Keep desirable writing style from masking factual errors.
Start with a capability hypothesis, then choose a second test with a different failure mode. Keep the checkpoint, reasoning effort, agent tools, sampling policy, and scoring procedure fixed when measuring a training intervention. Record failures and missing outputs in the denominator.
A reproducible comparison
Vision-arena preference rating reproduced in the Molmo author report. This dated vision cohort does not share a rating scale with text-only Arena or Molmo’s separate in-house human-preference study. The source confidence interval is retained in configuration.
Publish the exact run configuration, item coverage, score units, and available uncertainty. If a source supplies only an aggregate or a single run, preserve that limit. A new model name or a rounded score alone does not establish a capability improvement.
PRIMARY EVIDENCE & FURTHER READING
- Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models — published result table ↗
All 50 authors
Deitke, Matt, Clark, Christopher, Lee, Sangho, Tripathi, Rohun, Yang, Yue, Park, Jae Sung, Salehi, Mohammadreza, Muennighoff, Niklas, Lo, Kyle, Soldaini, Luca, Lu, Jiasen, Anderson, Taira, Bransom, Erin, Ehsani, Kiana, Ngo, Huong, Chen, YenSung, Patel, Ajay, Yatskar, Mark, Callison-Burch, Chris, Head, Andrew, Hendrix, Rose, Bastani, Favyen, VanderBilt, Eli, Lambert, Nathan, Chou, Yvonne, Chheda, Arnavi, Sparks, Jenna, Skjonsberg, Sam, Schmitz, Michael, Sarnat, Aaron, Bischoff, Byron, Walsh, Pete, Newell, Chris, Wolters, Piper, Gupta, Tanmay, Zeng, Kuo-Hao, Borchardt, Jon, Groeneveld, Dirk, Nam, Crystal, Lebrecht, Sophie, Wittlif, Caitlin, Schoenick, Carissa, Michel, Oscar, Krishna, Ranjay, Weihs, Luca, Smith, Noah A., Hajishirzi, Hannaneh, Girshick, Ross, Farhadi, Ali, Kembhavi, Aniruddha
- Chatbot Arena: An Open Platform for Evaluating LLMs by Human Preference ↗
All 11 authors
Chiang, Wei-Lin, Zheng, Lianmin, Sheng, Ying, Angelopoulos, Anastasios Nikolas, Li, Tianle, Li, Dacheng, Zhang, Hao, Zhu, Banghua, Jordan, Michael, Gonzalez, Joseph E., Stoica, Ion
Reviewed 2026-10-27. Explanation and interpretation by RLTasks. Benchmark ownership remains with its original authors.