RLTasksby ulam.ai
Models / OpenAI

OpenAI / MODEL EVIDENCE

GPT-3.5

15 benchmarks · 33 observations · 15 sources

Scores at a glance

15 benchmarks · large dot: GPT-3.5; small dots: closest models where reported

MuSR (Multistep Soft Reasoning) ↗61.6%
48.480.4

GPT-3.5 · MuSR (Multistep Soft Reasoning) · Accuracy · Murder mysteries · CoT+ ↗

2 more reported results

46.9% · GPT-3.5 · MuSR (Multistep Soft Reasoning) · Accuracy · Object placement · CoT+ · MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoning ↗

40.4% · GPT-3.5 · MuSR (Multistep Soft Reasoning) · Accuracy · Team allocation · CoT+ · MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoning ↗

JailbreakBench (JBB-Behaviors) ↗65%
065

GPT-3.5 · JailbreakBench (JBB-Behaviors) · JBB original-source attack success · source-subset comparison ↗

3 more reported results

47% · GPT-3.5 · JailbreakBench (JBB-Behaviors) · JBB original-source attack success · source-subset comparison · JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models ↗

0% · GPT-3.5 · JailbreakBench (JBB-Behaviors) · JBB original-source attack success · source-subset comparison · JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models ↗

87% · GPT-3.5 · JailbreakBench (JBB-Behaviors) · JBB original-source attack success · source-subset comparison · JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models ↗

WorkArena & WorkArena++ ↗6.1%
6.142.7

GPT-3.5 · WorkArena & WorkArena++ · WorkArena L1 · success rate · BrowserGym · tuned agent configurations ↗

BizBench ↗72.5 points
2884.6

GPT 3.5* / - · BizBench · Published task average · zero-shot and few-shot evaluation ↗

WildHallucinations ↗88.8%
75.0793.4

GPT-3.5 · WildHallucinations · WildFActScore · all entities · original evaluation ↗

BhasaBench / IndicGenBench ↗16.3 points
041.1

GPT-3.5 · BhasaBench / IndicGenBench · CrossSum-In · one-shot evaluation ↗

4 more reported results

29.2 points · GPT-3.5 · BhasaBench / IndicGenBench · Flores-In · one-shot evaluation · IndicGenBench: A Multilingual Benchmark to Evaluate Generation Capabilities of LLMs on Indic Languages ↗

33.2 points · GPT-3.5 · BhasaBench / IndicGenBench · XQuAD-In · one-shot evaluation · IndicGenBench: A Multilingual Benchmark to Evaluate Generation Capabilities of LLMs on Indic Languages ↗

21.6 points · GPT-3.5 · BhasaBench / IndicGenBench · XorQA-In-Xx · one-shot evaluation · IndicGenBench: A Multilingual Benchmark to Evaluate Generation Capabilities of LLMs on Indic Languages ↗

35.5 points · GPT-3.5 · BhasaBench / IndicGenBench · XorQA-In-En · one-shot evaluation · IndicGenBench: A Multilingual Benchmark to Evaluate Generation Capabilities of LLMs on Indic Languages ↗

WebArena ↗6.41%
6.416.41

GPT-3.5 · WebArena · End-to-end task success · original prompting configurations ↗

3 more reported results

8.75% · GPT-3.5 · WebArena · End-to-end task success · original prompting configurations · WebArena: A Realistic Web Environment for Building Autonomous Agents ↗

5.1% · GPT-3.5 · WebArena · End-to-end task success · original prompting configurations · WebArena: A Realistic Web Environment for Building Autonomous Agents ↗

6.16% · GPT-3.5 · WebArena · End-to-end task success · original prompting configurations · WebArena: A Realistic Web Environment for Building Autonomous Agents ↗

FreshQA ↗32.4%
32.432.4

GPT-3.5 · FreshQA · Relaxed with search · overall accuracy · source April 2023 snapshot ↗

Closest models by score

Similar positions on shared benchmark leaderboards. Smaller gaps are closer.

4.5 percentile-point gap

3 of 3 shared benchmark comparisons

How similarity is measured

Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.

Filter comparisons to this model →

Sources

Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.

  1. Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena — published result table ↗

    Zheng, Lianmin, Chiang, Wei-Lin, Sheng, Ying, Zhuang, Siyuan, Wu, Zhanghao, Zhuang, Yonghao et al. · 2023

    All 13 authors

    Zheng, Lianmin, Chiang, Wei-Lin, Sheng, Ying, Zhuang, Siyuan, Wu, Zhanghao, Zhuang, Yonghao, Lin, Zi, Li, Zhuohan, Li, Dacheng, Xing, Eric P., Zhang, Hao, Gonzalez, Joseph E., Stoica, Ion

  2. COLLIE: Systematic Construction of Constrained Text Generation Tasks — published result table ↗

    Shunyu Yao, Howard Chen, Austin W. Hanjie, Runzhe Yang, Karthik Narasimhan · 2023

  3. WebArena: A Realistic Web Environment for Building Autonomous Agents — published result table ↗

    Zhou, Shuyan, Xu, Frank F., Zhu, Hao, Zhou, Xuhui, Lo, Robert, Sridhar, Abishek et al. · 2023

    All 12 authors

    Zhou, Shuyan, Xu, Frank F., Zhu, Hao, Zhou, Xuhui, Lo, Robert, Sridhar, Abishek, Cheng, Xianyi, Ou, Tianyue, Bisk, Yonatan, Fried, Daniel, Alon, Uri, Neubig, Graham

  4. FreshLLMs: Refreshing Large Language Models with Search Engine Augmentation — published result table ↗

    Vu, Tu, Iyyer, Mohit, Wang, Xuezhi, Constant, Noah, Wei, Jerry, Wei, Jason et al. · 2023

    All 11 authors

    Vu, Tu, Iyyer, Mohit, Wang, Xuezhi, Constant, Noah, Wei, Jerry, Wei, Jason, Tar, Chris, Sung, Yun-Hsuan, Zhou, Denny, Le, Quoc, Luong, Thang

  5. MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoning — published result table ↗

    Zayne Sprague, Xi Ye, Kaj Bostrom, Swarat Chaudhuri, Greg Durrett · 2023

  6. BizBench: A Quantitative Reasoning Benchmark for Business and Finance — published result table ↗

    Rik Koncel-Kedziorski, Michael Krumdick, Viet Lai, Varshini Reddy, Charles Lovering, Chris Tanner · 2023

  7. VisualWebArena: Evaluating Multimodal Agents on Realistic Visual Web Tasks — published result table ↗

    Jing Yu Koh, Robert Lo, Lawrence Jang, Vikram Duvvur, Ming Chong Lim, Po-Yu Huang et al. · 2024

    All 10 authors

    Jing Yu Koh, Robert Lo, Lawrence Jang, Vikram Duvvur, Ming Chong Lim, Po-Yu Huang, Graham Neubig, Shuyan Zhou, Ruslan Salakhutdinov, Daniel Fried

  8. CriticBench: Benchmarking LLMs for Critique-Correct Reasoning — published result table ↗

    Zicheng Lin, Zhibin Gou, Tian Liang, Ruilin Luo, Haowei Liu, Yujiu Yang · 2024

  9. WorkArena: How Capable Are Web Agents at Solving Common Knowledge Work Tasks? — published result table ↗

    Drouin, Alexandre, Gasse, Maxime, Caccia, Massimo, Laradji, Issam H., Del Verme, Manuel, Marty, Tom et al. · 2024

    All 12 authors

    Drouin, Alexandre, Gasse, Maxime, Caccia, Massimo, Laradji, Issam H., Del Verme, Manuel, Marty, Tom, Boisvert, Léo, Thakkar, Megh, Cappart, Quentin, Vazquez, David, Chapados, Nicolas, Lacoste, Alexandre

  10. JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models — published result table ↗

    Patrick Chao, Edoardo Debenedetti, Alexander Robey, Maksym Andriushchenko, Francesco Croce, Vikash Sehwag et al. · 2024

    All 12 authors

    Patrick Chao, Edoardo Debenedetti, Alexander Robey, Maksym Andriushchenko, Francesco Croce, Vikash Sehwag, Edgar Dobriban, Nicolas Flammarion, George J. Pappas, Florian Tramer, Hamed Hassani, Eric Wong

  11. IndicGenBench: A Multilingual Benchmark to Evaluate Generation Capabilities of LLMs on Indic Languages — published result table ↗

    Harman Singh, Nitish Gupta, Shikhar Bharadwaj, Dinesh Tewari, Partha Talukdar · 2024

  12. WildHallucinations: Evaluating Long-form Factuality in LLMs with Real-World Entity Queries — published result table ↗

    Wenting Zhao, Tanya Goyal, Yu Ying Chiu, Liwei Jiang, Benjamin Newman, Abhilasha Ravichander et al. · 2024

    All 11 authors

    Wenting Zhao, Tanya Goyal, Yu Ying Chiu, Liwei Jiang, Benjamin Newman, Abhilasha Ravichander, Khyathi Chandu, Ronan Le Bras, Claire Cardie, Yuntian Deng, Yejin Choi

  13. GPQA Diamond — evaluation methodology and results ↗

    Vals AI · 2026

  14. LegalBench — evaluation methodology and results ↗

    Vals AI · 2026

  15. MedQA — evaluation methodology and results ↗

    Vals AI · 2026

Names used in source tables

GPT 3.5 · GPT 3.5* / - · GPT-3.5